Chromosomal Protein-DNA Interactions Citation Murphy, Robert Francis (1980) Chromosomal Protein-DNA Interactions. Dissertation (Ph.D.), California Institute of Technology. doi:10.7907/3ecg-z764. https://resolver.caltech.edu/CaltechTHESIS:09102025-214417021 Abstract The nature and function of some of the protein-DNA interactions in eukaryotic chromatin were investigated. The nucleosome structure of isolated template-active chromatin was determined. In vitro chemical acetylation of chromatin was shown to result in a structure similar to that of deproteinized DNA, which represents a shift towards the properties of isolated template active chromatin, and chromatin containing specific transcribed genes. The process of chromatin replication was shown to include a shortening of the internucleosomal spacer, resulting in decreased nuclease sensitivity. Newly-replicated chromatin was separated from bulk chromatin in shallow metrizamide density gradients. Newly-synthesized histone and newly-acetylated protein were shown to be present predominantly in the unreplicated chromatin fraction. The accuracy and reproducibility of non-linear least squares determinations of the thermal denaturation transitions of DNA and chromatin were determined using computer programs designed for ease of use and adaptability to mini-computer configurations. Direct fitting of melt data to a normalized error function gave results very similar to those obtained by fitting Gaussian curves to derivatized data. This approach avoids errors introduced by the derivatization method, and requires fewer data points. Item Type: Thesis (Dissertation (Ph.D.)) Subject Keywords: (Biochemistry) Degree Grantor: California Institute of Technology Division: Biology Major Option: Biochemistry Thesis Availability: Public (worldwide access) Research Advisor(s): Bonner, James Frederick Thesis Committee: Bonner, James Frederick (chair) Mitchell, Herschel K. Lazarides, Elias Strauss, James H. Richards, John H. Defense Date: 17 July 1978 Funders: Funding Agency Grant Number Weigle Fund UNSPECIFIED Record Number: CaltechTHESIS:09102025-214417021 Persistent URL: https://resolver.caltech.edu/CaltechTHESIS:09102025-214417021 DOI: 10.7907/3ecg-z764 ORCID: Author ORCID Murphy, Robert Francis 0000-0003-0358-901X Default Usage Policy: No commercial reproduction, distribution, display or performance rights in this work are provided. ID Code: 17676 Collection: CaltechTHESIS Deposited By: Benjamin Perez Deposited On: 12 Sep 2025 10:10 Last Modified: 12 Sep 2025 10:22 Thesis Files PDF - Final Version See Usage Policy. 25MB Repository Staff Only: item control page

CHROMOSOMAL PROTEIN-DNA INTERACTIONS Thesis by Robert Francis Murphy In Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy California Institute of Technology Pasadena, California 1980 (Submitted July 17, 1978) ii ACKNOWLEDGMENT To Dr. James Bonner, for providing me with the means and opportunity to study whatever seemed interesting; To Drs. Herschell Mitchell, Elias Lazarides, Jim Strauss, and Jack Richards, for agreeing to oversee this work; To Dr. R. Bruce Wallace, for his invaluable help and advice; To the members of the Bonner group, for their cooperation; To my teammates on various and diverse athletic teams, for making life in this one-horse town a little less unbearable; To Steve and Bonnie Petersen, for being my friends; And to my parents and family, with all my love. The preparation costs of this thesis were defrayed in part by the Weigle Fund. iii ABSTRACT The nature and function of some of the protein-DNA interactions in eukaryotic chromatin were investigated. The nucleosome structure of isolated template-active chromatin was determined. In vitro chemical acetylation o'f chromatin was shown to result in a struct_ure similar to that of deproteinized DNA, which represents a shift towards the properties of isolated template active chromatin, and chromatin ' containing specific transcribed genes. The process of chromatin replication was shown to include a shortening of the interriucleosomal spacer, resulting in decreased nuclease sensitivity. Newly-replicated chromatin was separated from bulk chromatin in shallow metrizamide density gradients. Newly-synthesized histone and newly-acetylated protein were shown to be present predominantly in the unreplicated chromatin fraction. The accuracy and reproducibility of pon-linear least squares determinations of the thermal denaturation transitions of DNA and chromatin were determined using computer programs designed for ease of use and adaptability to mini~computer configurations. Direct fitting of melt data to a normalized error function gave results very similar to those obtained by fitting Gaussian curves to derivatized data. This approach avoids errors introduced by the derivatization method, and requires fewer data points. iv CONTENTS PART I Chapter l. Alkaline extraction of non-histone proteins from rat liver chromatin 1 PART II Chapter 2. Structure of transcriptionally active chromatin 7 PART III Chapter 3. Physical properties of chemically acetylated rat liver chromatin 13 PART IV Chapter 4. Altered nucleosome spacing in newly replicated chromatin from Friend leukemia cells 19 Chapt e r 5. Properties of newly replicated chromatin: Analysis using shallow me trizamide gradients 25 PART V Chapter 6. Compute r programs for analysis of nucleic acid hybridization, thermal denaturation, and gel electrophoresis data 49 App endix A. GELFIT , a non-linear, least-squares Gaussian curve fitting program 62 Appendi x B. GELFIT fitting session example 84 1 PART I Chapter 1. Alkaline extraction of non-histone proteins from rat liver chromatin 2 Reprinted from Biochimica et Biophysica Acta, 405 (1975) 62-66 © Elsevier Scientific Publishing Company, Amsterdam - Printed in The Netherlands BBA 37139 ALKALINE EXTRACTION OF NON-HISTONE PROTEINS FROM RAT LIVER CHROMATIN ROBERT F. MURPHY and JAMES BONNER Division of Biology, California Institute of Technolog)l, Pasadena, Cal. 91 I 25 ( U.S.A.) (Received March 18th, 1975) SUMMARY The dissociation of non-histone proteins from chromatin by alkaline pH was s.tudied using sodium dodecyl sulfate-polyacrylamide gel electrophoresis. A significant degree of selectivity of dissociation was observed. This selectivity may provide a new method for fractionating non-histone proteins. INTRODUCTION Much work has been done on the selective dissociatiop of histones from nucleohistones at low pH [1-3], Other investigations have reported on the dissociation of histones from DNA at high pH, particularly at pH 12 arid above [4-7]. In this paper we describe experiments on the selective dissociation ofchromosomal proteins by titration in the range pH 8-11. MATERIALS AND METHODS Preparation of chromatin. Chromatin was prepared from Sprague-Dawley rat livers (Pel-freeze), by the sucrose purification method of Marushige and Bonner [8]. After pelleting through a I. 7 M sucrose cusion, the chromatin was resuspended in four volumes of 10 mM sodium bisulfite, 10 mM Tris buffer, pH 8.0, and spun at 12 000 x g for IO min. The chromatin pellet was resuspended at a concentration of 20 A 21,0 nm units per ml and sheared for 90 sat 30 Vin the Virtis homogenizer. The sheared chromatin was spun at I 2 000 x g for 20 min to remove a small amount of material not solubilized by shearing. Alkaline dissociation of chromatin. 10 ml of sheared chromatin was titrated to the desired pH (using a Radiometer 22 pH meter) by the dropwise addition of I M NaOH with stirring (pH values were checked by warming a small sample of the mat~rial to 25 °C and determining the pH again). The chromatin was then stirred for 30 min and centrifuged at 45 000 rev./min for 20 h in a Beckman Ti50 rotor ( 134 800 x g). The chemical composition of the material in the supernat.ants and pellets was determined, and the proteins of each sample were analyzed by polyacrylamide gel electrophoresis. Chemical composition. RNA and DNA were separated by the Schmidt- 3 63 Tannhauser procedure. The amount of DNA was determined from the absorption at 260 nm and RNA was determined by the orcinol procedure [9]. Total protein was determined by the procedure of Lowry et al. [10] using bovine serum albumin as standard. Histones were extracted with 0.2 M H 2 SO 4 and the insoluble non-histones removed by centrifugation at 12 000 x g for IO min. Pellets from this extraction were resuspended in I M NaOH and the amount of non-histone proteins determined by the procedure of Lowry et al. [JO], while the amount of histone protein was determined from the absorption at 230 nm using the relation I mg/ml = 4.15 A 230 nm units. Polyacrylamide gel electrophoresis. Sodium dodecyl sulfate-IO% acrylamide gels were run, stained and scanned as described previously, and the scans were resolved into a number of Gaussian curves suet, that the sum of those curves equaled the original data [I I, 12]. RESULTS Table I presents the results of protein and nucleic acid determination for the supematants and pellets of chromatins titrated to pH 8-11 and pelleted as described above. As can be seen, more than 95 % of the DNA is recovered in the pellets at all pH values. Increasing amounts of RNA, histone and non-histone are dissociated from chromatin as the pH is increased. The greatest increase in the release of all three components occurs between pH 10 and pH 11. This may be due to the titration of lysine residues (pK approx. 10.5). Indeed, in the titration of chromatin to pH 11, the buffering effect of the lysine residues is clear as evidenced by the need for a greater amount of NaOH to raise the pH from 10 to 11 than is needed to raise the pH from 8 to 9 or from 9 to 10. There is also a noticeable rise in viscosfty of the preparation above pH IO perhaps correlated with the dissociation of protein from DNA in some regions of the chromatin and concomitant unfolding of DNA. To discover whether particular non-histone proteins are released as the pH of chromatin is increased, we have run sodium dodecyl sulfate gels on the various fractions. Fig. I shows the composite gel scans of the supernatants (dissociated material) .at pH 8-11. Each Gaussian does not necessarily represent one peptide, but may represent one or more peptides of identical or similar molecular weight. TABLE I DISSOCIATION OF CHROMATIN CONSTITUENTS BY ALKALINE pH Chromatin isolation, titration and centrifugation were as described in Materials and Methods. Amounts are expressed as percent of starting material of each component ± one standard deviation. Ratio given·is the weight ratio of each component to DNA in whole chromatin . S, supernatant (dissociated); P, pellet (undissociated). pH RNA DNA s p 8 9 I I 99 10 11 2 2 Ratio: 1.0 99 98 98 .± s Histone p ± s 5 3 3 7 18 I I I 3 5 95 8 12 22 92 88 78 ± s p 0.5 0.7 0.8 0.8 12 26 36 88 74 9 64 JO 55 45 10 0.oJ I.I 97 97 93 82 Non-his tone p 0.8 ± 3 3 4 6 64 MOLECULAR WEIGHT 1451( 1801(11251( I I 66K I 37.5K 23K I I 16K I pH9 pHS ~ Fig. I. Composite gel scans of .protein dis30Ciatcd at several alkalinei'H values. Samples of 10011g protein from each supernatant were separated on 10-cm sodium do:lecyl .sulfate gols, and then the absorbances normalized to represent equivalent amounts of starting chromatin. Protein to DNA ratios arc as follows: pH 8, 0.07; pH 9, 0.09; pt-! ! 0, 0.17; pH 11, 0.35. In the first place there is a small general increase in release of material from chromatin as the pH is increased from 8 to 11. In addition, however, there is a significant increase in the release of particular protein fractions over this pH range. This increase in dissociated material is particularly marked in the high ( 140 ~ 180 000) molecular weight and low (22 000-43 000) molecular weight regions of the gel. Since mobility in these gels is proportional to log molecular weight, the molecular weight of each band can be determined using established standards (12]. The area under each peak is proportional to the mass of protein in that component. Therefore, the number of molecules in each band can be calculated. Table II presents the data in terms of molecules per genome (3 • 109 base pairs) and in percent of total number of molecules per genome for each molecular weight class [ 12). The fact that some classes of proteins are dissociated to a significantly higher degree or to a significantly lesser degree than the percent of total protein dissociated at that pH, is evidence of selectivity of dissociation. Proteins which are dissociated at pH 8 (the pH at which the chromatin is prepared) may be released by the shearing process (no control for this is possible since unsheared chromatin is an aggregate of very high molecular weight). In addition, the proteins dissociated at pH 8 evidently have a very low affinity for chromatin and may indeed be bound to chromosomal proteins rather than to DNA itself. 5 65 TABLE II DISSOCIATION OF NON-HISTONE PROTEINS BY ALKALINE pH R~ults of calculations on data in Fig. l. Total number of molecul~ column gives data from Garrard et al. [12) for numb« in each class per haploid genome (l.8 · 101 1 mol. wt). Molecular weight ( X 103) Total number of moiecules ( X 103) >195 171-190 150-170 140-149 130-139 100-129 92-98 80-90 68-79 46 101 122 92 91 554 31-43 23-30 < 22 122 450 1460 690 1005 4790 660 1950 1620 2876 6708 Total 22582 65 5fr63 50-55 48 45 Percent of total --· - --- pH 8 pH9 pH IO 13 1 2 0 0 3 0 1 2 0 0 1 17 51 51 5 19 9 0 0 106 35 64 0 2 0 0 6 pH 11 5 11 6 2 0 6 4 3 8 3 12 14 7 18 16 5 5 5 7 23 15 0 11 34 38 0 IO 41 62 0 13 14 70 98 28 5 8 12 22 0 3 4 5 5 One of the principal proteins released at pH 8 is a major non-histone and one which appears to be identical with the muscle protein, actin. It is possible that F actin is attached or bound to chromatin at one point and protrudes from chromatin as an F actin fiber, and is thus easily released by hydrodynamic forces. DISCUSSION The significance of our findings is 2-fold. First, we have established that the non-histone proteins of rat liver chromatin are subject to selective dissociation at alkaline pH . Second, our results indicate that it should be possible to enrich for a rather specific subset of non-histone chromosomal proteins by first titrating chromatin to pH 10, removing the dissociated material and then titrating the chromatin again to pH 11 and removing the dissociated material. This procedure would provide still another method of fractionating and characterizing the non-histone chromosomal proteins. ACKNOWLEDGMENTS We wish to thank Dr Angeline Douvas, Mr William R. Pearson and Mr Joel Gottesfeld for advice and helpful counsel, as well as Ms Ryn Miake for technical assistance. 66 REFERENCES 1 Johns, E. W. and Butler, J. A. V. (1962) Biochem. J. 82, 15-18 2 Murray, K. (1966) J. Mol. Biol. 15, 409-419 3 Murray, K., Vidali, G. and Neelin, J. M. (1968) Biochem. J. 107, 207-215 4 Matsuyama, A., Tagashita, Y. and Nagata, C. (1971) Biochim. Biophys. Acta 240, 184-190 5 Ramm, E. I., Vorob'ev, V. I., Birshtein, T. M., Bolotina, I. A. and Volkenshtein, M. K. (1972) Eur. J . Biochem. 25, 245-253 6 Ohba, Y. and Hayashi, M. (1972) Eur. J. Biochem. 29, 461-468 7 Russev, G., Venkov, C. and Tsavev, R. (1974) Eur. J. Biochem. 43, 253-256 8 Marushige, K. and Bonner, J. (1966) J. Mo!. Biol. 15, 160-174 9 Dische, Z. and Schwartz, K. (1937) Mikrochim. Acta 2, 13 10 Lowry, 0. H., Rosebrough, N . J., Farr, A. L. and Randall, R . J. (1951) J. Biol. Chem. 193, 265275 11 King, J. and Laemmli, V. K. (1971) J. Mol. Biol. 62, 465-477 12 Garrard, W. T., Pearson, W. R., Wake, S. K. and Bonner, J. (1974) Biochem. Biophys. Res. Commun. 58, 50-57 7 PART II Chapter 2. Structure of transcriptionally active chromatin 8 Reprinted from Proc. Nat . Acad. Sci. USA Vol. 72, No. 11 , pp. 4404- 4408, November 1975 Biochemistry Structure of transcriptionally active chromatin (chromatin fractionatioa/ deoiyriboauclease II/chromatin subuoits/ ... bodies) JOEL M . GOTTESFELD•, ROBERT F. MURPHY, AND JAMES BoNNER Division of Biology, California Institute of Technology , Pasadena. Calif. 91125 Contributed by Jame, Bonner, September IO, 1975 clease-resistant structures of active chromatin are complexes of DNA, RNA, histone, and nonhistone chromosomal protein. ABSTRACT Rat-liver chromatin bas been fractionated into transcriptionally active and inactive regions [Cottesfeld et a/. (1974) Proc. Nat. A.cad. Sci. USA 71, 2193--2197) and the distribution of nuclease-resistant comflexes in these fractions has been investigated. About hal of the DNA of both fractions is resistant to attack by the endonuclease DNase II. The nuclease-resistant structures of inactive chromatin are DNA-histone complexes (.,.bodiesl which sediment at 11-13 S. Template-active chromatin yie ds two peaks of nucleaseresistant nucleoprotein. These complexes sediment at 14 and 19 S, and contain DNA, RNA, histone, and nonbistone chromosomal proteins. Polyacrylamide gel electrophoresis reveals a complex pattern of chromatin proteins, suggesting that the complexes are heterogeneous in composition. MATERIALS AND METIIODS Chromatin Fractionation. Chromatin was prepared from rat liver by the method of Marushige and Bonner (22) and treated with diisopropylfluorophosphate (DFP) to inhibit endogenous protease activity (23). Fractionation was carried out as diagrammed in Fig. l ; details of this method have been published previously (21). P'ttparation of Chromatin Subunits. Nuclease-resistant subunits of rat-Ii ver chromatin were prepared as follows: DNase II was added to 10 units per A2eo unit of chromatin in 25 mM sodium acetate (pH 6.6). Digestion was carried out at 24° and was terminated after 90 min by raising the pH to 7.5 with 0.1 M Tris•HCl (pH 11). Nuclease-resistant subunits from chromatin fraction Pl were prepared by homogenizing the pellet fraction in 25 mM sodium acetate (pH 6.6) and redigesting with DNase as described above for whole chromatin. Undigested chromatin (about 20% 0£ the input DNA) was removed b~ centrifugation at 27,000 X g for 10-15 min. The supe.J"'tant was layered on isokinetic sucrose gradients in SW25.l cellulose nitrate tubes. The gradients were formed according to Noll (24); the parameters were Crop= 15% (weight/ volume), CRES = 34.2% (weight/ volume), and VM1x = 31.4 ml. All solutions contained 10 mM Tris-HCI (pH 8). Centrifugation was at 25,000 rpm for 36-42 hr. Gradients were analyzed with an ISCO UV Analyzer and chart recorder. Fractions from these gradients were rerun on 5-24% isokinetic sucrose gradients. The parameters were CToP = 5.1% (weight / volume), GRES = 31.4% (weight/ volume) and VMIX = 9.4 ml. Centrifugation was in the SW 41 rotor at 39,000 rpm at 4° for 16.5 hr. Subunits were also prepared from chromatin devoid of histone I. Removal of this histone was accomplished by extraction of Virtis-sheared chromatin (45 V, 90 sec) with 0.5 M NaCl at 4°. The resultant nucleohistone was pelleted by centrifugation in the Ti 50 rotor at 50,000 rpm for 18 hr. The pellet was digested with nuclease as described above. Redigestion of Chromatin Fraction S2. Chromatin of fraction S2 was redigested with nuclease in three different ways: the DNase II present in fraction S2 from the first nuclease treatment was reactivated by adding EDTA to 20 mM and lowering pH to 6.4 with dilute HCl. Alternatively, akquots of chromatin fraction S2 were dialyzed against either 25 mM ,,odium acetate (pH 6.6) or 5 mM sodium phosphate (pH 6.7) containing 2.5 X 10-• M CaCh and 2.5 X 10- ◄ M MgCl 2 . DNase lI was added to the chromatin in sodium acetate buffer to 10 units/ A26Cl unit of chromatin; staphylococcal nuclease was added to the chromatin in sodium phosphate buffer to 50 units/ ml. Reactions were carried out at A regular repeating unit in chromatin was first suggested from the x-ray diffraction studies of Pardon et al. (1): a series of reflections were observed in the • -ray patterns of native and reconstituted nucleohistones, but not in the x-ray diffraction patterns of DNA or histones by themselves. It was proposed that the chromatin fiber is organized into a regular supercoil of pitch 100-120 A. This model, although widely accepted for some time, has now come under question. Olins and Olins (2) h·a ve observed regular spacings of chromatin particles (termed v-bodies) in water-swollen nuclei centrifuged onto electron microscope grids. These parti• des are 60-80 A in diameter and are joined by thin filaments 15 A in diameter. These results have been confirmed and extended by other laboratories (3-5). Nuclease digestion studies also support a subunit or particulate structure for chromatin. Both endogenous (6-7) and exogenous (8-14) nu·cleases appear to recognize a repeating nucleoprotein unit along the chromatin fiber. Furthermore, chromatin particles isolated from nuclease-treated or sonicated chromatin resemble ,-bodies in the electron microscope (15-17). Thus many lines of evidence support the subunit or "beads-on-astring" model of chromatin structure (2, 18). Studies on chromatin structure have been generally carried out with whole chromatin, unfractionated with respect to transcriptional activity. Since only a minor portion of the DNA in any differentiated cell type is ever transcribed into RNA, the properties of unfractionated chromatin reflect primarily the structure of inactive regions. We are interested in whether template-active chromatin is organized as is the inactive region or whether it is in a different conformation. Previous worlr. from this laboratory has shown that it is possible to separate chromatin into transcriptionally active and inactive fractions (19-21 ). In this communication we report that both fractions contain nuclease-resistant nucleoprotein complexes. However, the nuclease-resistant structures of inactive chromatin are ONA-histone complexes, while the nuAbbreviation: DFP, diisopropylfluorophosphate. • Present address: MRC Laboratory of Molecular Biology, Hills Road, Cambridge, England CB2 2QH. 4404 9 Biochemistry: Gottesfeld el al. 24° . Aliquots were taken at various times to test for the production of trichloroacetic-acid-soluble material (measured by absorbance of the supernatant at 260 nm after centrifugation at 27,000 X g at 4° for 15 min). DNA Size Estimation. Single-strand DNA lengths were estimated by velocity sedim,•ntation in all<aline sucrose gradients. The parameters for the isokinetic gradients were Cmp = 15.9"1 (weight/ volume), GRES= 38.9% (weight/ volume), and VMIX = 6.1 ml (24). All solutions contained 0.1 N NaOH. Chromatin samples were suspended in 0. 1 N NaOH. 2% sodium dodecyl sulfate, 2 M urea, and 100-200 µI aliquots were layered on each gradient. Centrifugation was in the SW50.l rotor at 48,000 rpm for 16 hr at 20°. DNA molecular weights were determined relative to a standard sized by electron microscopy (320 nucleotide-long, 5.4 S, calf thymus DNA; a gift of Ms. M. Chamberlin). Double-strand lengths were determined in the analytical ultracentrifuge (25). Analysis of Chromatin Composition. Histone and nonhistone protein content was determined as described (26). Protein was analyzed by sodium dodecyl sulfate-disc gel electrophoresis (27) and by acid-urea gel electrophoresis (28). DNA and RNA were determined by the methods of Schmidt and Tannhauser (29). Enzymes. DNase II (EC 3.1.4.6) and micrococcal nuclease (EC 3.1.4.7) from Staphylococcus were purchased from Worthington. RESULTS Chromatin Fractionation. DNase II preferentially attacks a select portion of chromatin DNA. The amount of DNA in this fraction varies depending upon the source of the chromatin, but corresponds quite closely to the measured template activity of the particular chromatin (20). The fractionation scheme used herein is diagrammed in Fig. I. After 5 min exposure to DNase II, 15% of rat-liver chromatin DNA remains soluble after centrifugation (fraction Sl). About 11% of the total DNA is Mg++.soluble and is found in fraction S2. This DNA comprises a subset of whole genomal DNA sequences and is enriched 5-fold in transcriptionally active sequences (21). The DNA has a double-strand length of about 700 base pairs and a single-strand length of 200600 nucleotides (range of observed values). About 1-3% of this DNA is acid-soluble. After 30 min exposure to DNase II, nearly 80% of the chromatin is found in fraction SI , and 20-24% is found in fraction S2. After prolonged nuclease digestion roughly half of the DNA of both fraction SI and fraction S2 is acid-soluble. A more detailed description of the kinetics of DNase II action on chromatin is presented elsewhere (J. Gottesfeld, G. Bagi, B. Berg, and J. Bonner, manuscript submitted). Table l lists some of the properties of the chromatin fractions: the composition of fraction Pl is similar to that of unfractionated rat-liver chromatin (21, 22). Fraction S2, however, is enriched in RNA and nonhistone protein and has a reduced content of histone protein. All the major histone species are present in fraction S2 isolated from DFP-treated chromatin. In a previous communication (21) we reported that fraction S2 lacked histone I; the absence of this histone was presumable due to its proteolytic degradation, since the chromatin was not treated with protease inhibitors. Subunit Structure of Chromatin. Rat-liver chromatin and fraction Pl chromatin have been digested with DNase II for extended periods of time (90 min), and the resulting soluble chromatin has been analyzed by centrifugation in Proc. Nat. Acad. Sci. USA 72 /1975) 44().5 Chromatin (370 J,,1-0 / ml ONA, ONose n ( 10 0 uni1s/ml} l 2 5 mM No Acetate, pH 6 .6) lnc.ubott ~ min a t z 4• Centrlfuo• (27,000 9 , l~mln) Petitt Supernatant 0 0 l 84 .6 ! 4 .8% 2 mM Moc1 2 ot ONA Centrltuo• (27,000o, 15 m in) I Pellet S upernatant 4 . 1 ! 2. 5% 11. 3 ! 3.9% e 0 FIG. I. Fractionation scheme. The yields of DNA in each frac tion are the mean and standard deviation for 11 determinations. isokinetic sucrose gradients (Fig. 2). About 40% of the input DNA sediments extremely slowly; the bulk of this DNA is acid-soluble and hence has been reduced to oligonucleotides by the nuclease. Fig. 2 prl!sents data for whole chromatin depleted of histone I (Fig. 2A) and for Pl chromatin (Fig. 2B). Most of the chromatin sediments at about 11-13 S, with some material sedimenting more rapidly. The properties of nuclease-resistant particles from unfractionated chromatin have been described in detail elsewhere (8-14, 16, 17). We find similar properties for the particles from PI chromatin: they are composed of equal amounts of protein and DNA (by weight), and the protein complement is almost entirely histone (Table I). Subunits from native chromatin sediment slightly more rapidly than suqJri,its from histone-1-depleted chromatin (Fig. 2). The sedimentation coefficients (± SD) estimated from ten isokinetic 5-24% sucrose gradients were 12.6 ± 0.9 S for subunits of native chromatin and 11.4 ± 0.7 S for subunits of histone-1-depleted chromatin. The length of DNA contained in the subunits has been studied by many workers; values of 120 to 210 base pairs have been obtained Table 1. Chemical composition of rat-liver chromatin fractions Composition relative to DNA (w / w) NonSample Histone histone RNA 1.06 1.15 1.03 0.61 0.65 0.58 < 0.05 1.60 0.05 0.05 0.24 0.72 0.54 0.60 1.35 3.2 Unfractionat.!d chromatin Pl chromatin* 1i-13S subunitst S2 chromatin• 0.25 S2 subfractions 3-5 S 14.0 , o.sst 18.1 , 1.ost 0.3-0.4 0.3-0 .7 • Fractionation carried out as in Fig. 1. t Similar compositions were obtained for native and histone -1- depleted subunits. t Sedimentation values ± SD determined from 24 gradients equivalent to those in Fig . 3. 10 Biochemistry: Gottesfeld et al. 4406 Proc. Nat . Acad. Sci. USA 72 (1975) 60 e < I 8 < ·' 10 I 16 Too FIG. Fraction Number 2. 2•1 I 10 Botlan Too Frac tion Number 00 0 15 40 80 20 40 60 80 120 16 0 (4 0 ) (0 ) Tme (IN"I) Bottom Sucr0&e gradient sedimentation of DNase-11 -treated chromatin samples. (A) Histone•l -depleted chromatin was digested and soluble chromatin was centrifuged for 42 hr as discussed. (8) Fraction Pl chromatin wa., digested and soluble chromatin was centrifuged for 36 hr. Fractions were pooled as indicated and rerun on 5-24% gradients (insets) . (10. 13, 14, 16, 17). The 11- 135 subunits resemble ,-bodies (2) in the electron microscope ( 17). Subunit Structure of Active Chromatin. We now ask · whether nuclease-resistant structures occur in transcriptionally active regions of chromatin. Chromatin from rat liver was treated with DNase II for 5 min, fractionated as before (Fig. 1). and S2 material was analyzed on isokinetic sucrose gradients (Fig. 3, curve A). About half of the UV-absorbing material applied to the gradient sediments at 3-5 S; greater than 90% of this material is acid-precipitable after the 5 min nuclease treatment. Two more rapidly sedimenting peaks are seen in the gradient of S2 chromatin, one at 14 S and another at 19-20 S. These gradients were calculated for particles of density 1.44 g/ cm 3 , and so the observed sedimentation coefficients could be in error if the particle densities are very different from 1.4 to 1.5 g/ cm 3 . About 6% of the input nucleic acid pelleted during the centrifugation. To test whether the 14 and 19S nucleoprotein complexes might be multimers of the 11-135 subunit (i.e., dimers and trimers), chromatin fraction S2 was reincubated with DNase. Upon analysis in isokinetic sucrose gradients (Fig. 3. FIG. 4. Kinetics of digestion of chromatin fraction S2. Chro- matin was fractionated as described (Fig. 1). S2 chromatin was incubated with nuclease& in three ways: reactivation of DNase II (0); addition of fresh DNase II (A) ; addition of staphylococcal nuclease (□) . Aliquots were taken at various times to test for the production of trichloroacetic-acid-soluble material. curve B) no significant changes were observed in the > lOS region. Reincubation with DNase has been carried out in three ways (reactivation of DNase II, addition of fresh DNase II, addition of staphylococcal nuclease), and similar results were obtained with each of these methods. The kinetics of redigestion of S2 chromatin are presented in Fig. 4. With each method of redigestion. about 50-60% of the input A200 became acid-soluble. As the reaction approached completion the solutions first became turbid and eventually a precipitate developed. S2 chromatin has been isolated after various times of nuclease treatment {30 sec to 30 min) and analyzed in sucrose MOLECULAR ~EIGHT If't :r1 l11m11 ~ ,·nn tD ,-.. LA .II m 1-- 11 r'I II N 3 - 55 l 1., "' N c "u C 0 .c ~ .c "' Top FIG. 3. Sucrose gradient sedimentation of template-active fraction S2 chromatin. Curve A: chromatin was fracti onated (Fig. I) and S2 material was centrifuged for 17.S hr at 39,000 rpm in 5-24% sucrose gradient. Curve B: fraction S2 was isolated and DNase II reactivated by the addition of 20 mM EDTA (pH 6.4). Incubation was for 1 hr at 24° . The reaction was terminated by raising the pH to 8 with 0.1 M Tris-HCI (pH 11), and the sample was centrifuged as described above. FIG. 5. Sodium dodecyl sulfat.e--polyacrylamide gels of S2 chromatin subfractiona. Total protein of the 3-5S (upper gel) and 14S (lower gel) complexes waa electrophoresed as described (27) . The stained gels were scanned and the densitometer profiles were resolved into gaussian components by a least-squares computer analysia. 11 Biochemistry: Gottesfeld et al. gradients. It was found that the 14 and 19S complexes appear in fraction S2 simultaneously. Thus there is no evidence for a precursor-p_roduct relationship between the 19 and 14S complexes. The nuclease-resistant S2 DNA (14 and 19S) was found to have a weight-average single-strand length of l 70 nucleotides after the initial DNase treatment (5 min). After prolonged digestion (up to 120 min), the single-strand length was reduced to 120 nucleotides. Nearly all the DNA which sedimented at 3-5 S after the initial DNase digestion was rendered acid-soluble by redigestion. We have investigated the chemical compositions of the subfractions of S2 chromatin (Table l). The material at 14-19 S is enriched in both RNA and nonhistone chromosomal proteins. All the major histone species are present in the 14 and 19S complexes; however. these complexes exhibit reduced histone to DNA ratios relative to either unfractionated chromatin or ll - 13S subunits (Table l). The protein population of S2 chromatin has been investigated by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (Fig. 5). Material from the 3-5S (upper gel) and from the 14S (lower gel) position in the sucrose gradient (Fig. 3) has been analyzed. Densitometer scans of the stained gels were resolved into gaussian components by a least squares computer analysis. Numerous quantitative and qualitative differences between the 3-5S and l 4S proteins can be recognized in these gels. From the length of DNA contained in the 14S chromatin complex (170 nucleotides). the protein-to-DNA ratio (Table l ), and the complexity of the protein population (Fig. 5). we conclude that the 14S peak consists of a heterogeneous population of nucleoprotein species. DISCUSSION The template-active fraction of rat-liver chromatin is organized in a fashion similar to that of inactive chromatin; both fractions consist of regions of nuclease-sensitive and nuclease-resistant DNA. Nuclease-resistant segments in transcriptionally inactive chromatin are due to histone-DNA interactions, while the nuclease-resistant segments of active chromatin are due to DNA complexed with both histone and nonhistone proteins. Nuclease-resistant structures of inactive chromatin sediment at 11-13 Sand resemble v-bodies (2) in the electron microscope ( 17). The nuclease-resistant complexes of active chromatin sediment at 14-19 S and contain RNA as well as protein and DNA. Our results shed new light on the findings of Felsenfeld's laboratory (8, 9, 31). These investigators have reported that portions of the globin gene are found in both nuclease-sensitive ("open") and nuclease-resistant ("closed") regions of reticulocyte chromatin. Our results suggest that nuclease sensitivity in a limit digest does not discriminate between active and inactive chromatin regions. Felsenfeld's data indicate that regions of the globin genes are always covered with protein (30) but make no distinction betw~n histone and nonhistone protein. On the basis of our data we speculate that active genes are complexed with nonhistone as well as histone protein in the 14 and l 9S structures. Inactive genes are complexed mainly with histone and are in the v-body structures. Although active, like inactive, chromatin contains nuclease-resistant and sensitive regions, there are other major differences. Thermal denaturation and circular dichroism studies. (31-33) suggest that active chromatin is in a more extended, more DNA-like conformation than inactive chromatin. The electron microscope has revealed differences in the structure of transcriptionally active and Proc. Nat . Acad. Sci. USA 72 (1975) 4407 inactive regions of chromatin. Ribosomal genes in the act of transcription are the length of their transcription product (pre-rRNA) (34). On the other hand, DNA complexed with histones in the ... body configuration is one-seventh the length of the same deproteinized DNA (4). The basic fiber diameter of inactive chromatin is 100 A. Active chromatin has a fiber diameter of about 30 A (35, 36). Thus both physical chemical and electron microscopy studies suggest that DNA of active chromatin is more extended than is the DNA of inactive chromatin. It is probable that this is why active chromatin is more susceptible to nuclease attack. This differential sensitivity to nuclease forms the basis of our fractionation procedure (Fig. 1). We wish to thank Dr. K. Peten and Mr. W. Wheatley for helpful discussions. This work was supported by the U.S. Public Health Service (GM 86 and GM 13762). I. 2. 3. 4. 5. 6. 7. 8. 9. 10. II. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. 25. 26. 27. 28. Pardon, J. F .. Wilkins, M. H. F. & Richards, B. M. (1967 ) Na ture 215, 50S-509. Olins, A. L. & Olins, D. E. (1974) Science 183, 330-332. Woodcock, C. L. F. (1973)]. Cell Biol. 59, 368a. Griffith, J. D. (1975) Science 187, 1202- 1203. Oudet, P., Gross-Bellard, M. & Chambon, P (1975) Cell 4, 281-300. Hewish, D. R. & Burgoyne, L. A. (1973) Btochem. Blophys. Re,. Commun. 52, 504--510. Burgoyne, L.A. Hewish, D. R. & Mobbs, J. (1974) Blochem. ]. 143, 67- 72. Clark, R. & Felsenfeld, G. (1971) Nature 229, 101 - 106. Clark, R. & Felsenfeld. G. (1974) Biochemistry 13, 36223628. Axel, R. , Melchior, W., Jr., Sollner-Webb, B. & Felsenfeld, G. (1974) Proc. Nat . Acad. Sci. USA 11, 4101 - 4100. Rill, R. & van Holde, K. E. (1973)]. Biol. Chem . 248, 10801083. Sahasrabuddhe, C. G. & van Holde, K. E. (1974) ]. Biol. Chem. 249, 152- 156. Noll, M. (1974) Nature 251, 249-252. Oosterhof, D. K.. Honer, J. C. & Rill. R. L. (1975) Proc. Nat . Acad. Sci. USA 72, 633-637. van Holde, K. E.. Sahasrabuddhe, C. G., Shaw, B., van Brugger, E. F. J. & Amberg, A. C. (1974) Blochem. Blophys. Re.s. Commun. 60, 1365-1370. Senior, M. B., Olins, A. L. & Olins, D. E. (1975) Science 187, 17l-175. Gottesfeld, J. M., Kent, D., Rou, M. & Bonner, J. (1975 ) in Florida Colloquium on Molecular Biology, eds. Stein, G. & Stein, J. (Academic Press, New York). in press. Kornberg, R. (1974) Science 184, 86S-87I. Marushige, K. & Bonner, J. (1971) Proc. Nat . Acad. Sci. USA 68, 2941-2944. . Billing, R. J. & Bonner, J. (1972) Blochlm. Blophfll. Acta 281, 453-462. Gottesfeld, J. M.. Garrard, W. T.. Bagi, G.. Wilson, R. F. & Bonner, J. (1974) Proc. Nat . Acad. Sci. USA 71, 2193-2197. Marushige, K. & Bonner, J. (1966)]. Mo/. Biol. 15, 160-174. Douvas, A., Harrington, C. & Bonner, J. (1975) Proc. Nat . Acad. Sci . USA 72, in press. Noll, H. (1967) Nature 215, 360-363. Studier, F. W. (1965)]. Mo/. Bwl. II , 373-390. Bonner, J.. Chalkley, R., D_a hmw, M.. Fambrough, D., Fujimura, F ., Huang. R. C. C. , Huberman, J.. Jensen, R., Marushige, K.. Olenbusch. H .. Olivera, B. & Widholm. J. (1968) in Method, In Eru:ym,,/ogy, eds. Grossman, L. & Moldave, Ii::. (Academic Press, New York), Vol. 12, part B, pp. l-65. King, J. & Laemmli, U. K. (1971)]. Mo/. Biol. 62, 465-477. Panyim, S. & Chalkley, R. (1969) Arch. Blochem. Blophys. 130, 337-346. 12 4408 29. Proc. Nat . Acad. Sci. USA 72 (1975) Biochemistry: Gottesfeld et al. Schmidt, G. & Tannhauser, S. J. (1945)]. Biol. Chem. 161, 35. Polacow, I. & Simpson, R. T. (1973) Bwchem. Bwph!/0, Re,. Commun. 52, 202- 207. Miller, 0 . L. & Hamkalo, B. A. ( 1972) in Molecu/4r GmeUc, and Developmental Biology, ed. Sussman, M . (Prentice-Hall, Englewood Cliffs, N.J.), pp. 183- 199. Griffith, J. (1970) Ph.D. Dissertation, California Institute of 36. Technology, Pasadena, Calif. Miller, 0 . L. ( 1965) Na t. Cancer Inst. Monogr . 18, 79- 89. 33. 83- 89. 30. 31. 32. Axel, R. , Cedar, H. & Felsenfeld, G. (1973) Cold Spnng Harbor Symp. Quant. Biol. 38, 773- 783. McConaughy, B. L. & McCarthy, B. J. (1972) Bwchemutry 11 , 99S-1003. Gottesfeld, J. M., Bonner. J., Radda, G. K. & W alker, 1.0. (1974) Bwchemutry 13, 2937 - 2945. 34. 13 PART III Chapter 3 . Physical properties of chemically acetylated rat liver chromatin 14 Reprinted from Proc. Natl . Acad. Sd. USA Vol. 74, No. 8, pp. 3244- 3248 , August 1977 Biochemistry Physical properties of chemically acetylated rat liver chromatin (acetic anhydride / nucleosomN/ thenruu denaturation/ histones) R. BRUCE WALLAC E, THOMAS 0 . SARGENT, ROBERT F. MURPHY, AND JAMES BONNER Divuion of Biology. California Institute 0£ Trchnology, Pasadena, California 91125 Contributed by Jame, Bonner, May 23, 1977 ABSTRACT The physical properties of rat liver chromatin and nucleosomes acetylated with acetic anhydride were examined in order to clarify the mechanism by which chemical acetylation of histones increases template activity in vitro [Marushige, K. (1976) Proc. Natl. A.cad. Sci. USA 73, 3937-3941[. Acetylation was found to have dramatic effects on the magne· sium solubility, nuclease sensitivity, thermal denaturation, and sedimentation of chromatin and nucleosomes. The significance of the results to models of gene activation and chromatin replication is considered. The five principal species of histone are associated with DNA to form the eukaryotic chromosome. These basic proteins are involved in the repression of template activity of chromatin (I , 2). It has long been thought that a degree of transcriptional control might be achieved by the selective modification of certain basic residues of the histones (3, 4). In the current model of chromatin structure, 200 base pairs of DNA are complexed with histones to form the nucleosomal subunit. Although the nucleosome is a general feature of all eukaryotic chromatin thus far examined, there appears to be heterogeneity within the nucleosomes of any single cell type (5). Some of this heterogeneity consists of different conformations of nucleosomes of active and inactive genes (6, 7). which may result from histone modification (4). Marushige (8) has reported that chemical acetylation of calf thymus chromatin increases its template activity In Pitro without resulting in significant removal of the histones from the DNA. These results suggest that acetylation of histones participates in activation of genes for transcription. In this report we describe the effect of chemical acetylation on the physical properties of rat liver chromatin and nucleosomes. MATERIALS AND METHODS Preparation of Nuclei. Rat liver nuclei were prepared from frozen rat ltvers (Pelfreeze). All manipulations were performed at 0° unless otherwise stated. Livers were thawed in 0.25 M sucrose/ 10 mM Tris•HCI, pH 7.4 / 6 mM KCl / 5 mM Mg(OAch/ 0.l mM ethylene glycol-bis(/l-aminoethyl ether)· N-N'-tetraacetic acid (EGTA). Livers were transferred to three volumes of the same buffer containing I mM phenylmethylsulfonyl fluoride (homogenization buffer) and homogenized with five to seven strokes in a motor-driven glass/ Teflon homogenizer. The homogenate was diluted to 10 volumes with homogenization buffer, strained through cheesecloth, and centrifuged at 1500 rpm in an HG-4L rotor (Sorvall RC-3 centrifuge) for 10 min. The pellet was washed twice in 10 volumes of homogenization buffer containing 0.5% Nonidet-P40 (Particle Data Laboratories, Ltd ), two to four times in homogeni7.ation buffer containing 0.1 mM phenylmethylsulfonyl The coots of publication of this article. were defrayed in part by the payment of page charges from funds made available to support the research which is the subject of the article. This article must therefore by hereby marked "advertisement" in accordance with 18 U. S. C. § 1734 solely to indicate this fact. fluoride and 0.5% Nonidet-P40, and finally once in homogenization buffer. The nuclear pellet was resuspended in homogenization buffer and an equal volume of 5 mM Mg(OAc)z/ 98% glycerol was added. This suspension is stored at -20° for up to I month. Preparation or Chromatin. Nuclei were recovered from the glycerol suspension by centrifugation at 3000 rpm in an HB-4 rotor (Sorvall RC-2 centrifuge) and washed once in 10 mM Tris-HCI, pH 7.4/6 mM KCl/ 5 mM Mg(OAc)z (TKM). The nuclei were lysed by vigorous homogenization in 2.5 mM Tris-HCI, pH 7.4/ 2.5 mM ethylenediaminetetraacetic acid (EDTA), followed by centrifugation at 10,000 rpm in the HB-4 rotor for 10 min. The chromatin was then washed once in the same buffer and twice in 5 mM sodium borate, pH 8.2/ 10 mM NaCl and finally was suspended in 5 mM sodium borate, pH 8.2/ 10 mM NaCl at an A250 of 10 (all chromatin absorbances are measured in I M NaOH). Novikoff hepatoma cells !line NISl-67 adapted for growth in Swimms 210 medium (Gibco)] were cultured as described by Plagemann (9). Cells were labeled with l2 - 14CJthymidine (56 mCi/mmol) at 0.05 µCi/ml for two cell generations. Preparation of nuclei will be described elsewhere (R. B. Wallace, S. K. Dube, and J. Bonner; unpublished data); they were mixed with 4-fold excess of rat liver nuclei . Chromatin was prepared as described above. Preparation of Nucleosomes. Nuclei were recovered from the glycerol suspension as described above, washed once in TKM, and suspended in TKM at an A260 of 150-200. The suspension was warmed to 37°, brought to 0.25 mM CaC'2. and digested for 5 min with 0.24 µg (7 .5 units) of staphylococcal nuclease (P-L Biochemicals) per A250 unit. The nuclei were cooled to 0° and centrifuged at 3000 rpm in the HB-4 rotor for 5 min. The nuclear pellet was homogenized in 2.5 mM Tris· HCI, pH 7.4/ 2.5 mM EDT A and centrifuged at 10,000 rpm in the HB-4 rotor for 10 min. The supernatant was then passed over a Sephadex G-50 column equilib.ated with 5 mM sodium borate, pH 8.2/0.l mM EGTA; the excluded fraction, representing nucleosomes, was adjusted to an A250 of 10. Typical preparations were 5-15% acid soluble and 50-70% monomer nucleosomes. Acetylation or Chromatin and Nucleosomes. Chromatin and nucleosomes were acetylated with acetic anhydride as described by Wong and Marushige (IO). Chromatin solutions were acetylated with 0.14 mM, 0.7 mM. and 7 mM acetic anhydride at an A260 of 10. Acetylated chromatin samples were dialyzed overnight against 10 mM Tris-HCI, pH 7.4. The incorporation of acetate groups was monitored by acetylating chromatin with !3H}acetic anhydride (Schwartz/ Mann) diluted to 0.4 mCi/mmol. The amount of acetylation achieved with the three concentrations was essentially the same as that reported by Wong and Marushige. Abbreviations: EGTA, ethylene glycol-bis(/l-aminoethyl ether)• N,N' -tetraacetic acid; TKM buffer, 10 mM Tris•HCI, pH 7.4/ 6 mM KCl/ 5 mM Mg(OAc).: EDTA, ethylenediaminetetraacetic acid. 3244 15 Aiochemistry: Wallace et al. Proc. Natl. Acad. Sci. USA 74 (1977) 3245 100 0 0 75 25 10 20 30 0+---~- -~-~"" 0 20 40 60 Time. min + 25 • • • 0 - t - - - ~ - - ~ -- - ~--~--...,......J 0 4 6 8 Magnesium aceta te, mM FIG. I. Solubility of acetylated chromatin in magnesium acetate. Nucleooomes were acetylated with acetic anhydride at O( ♦ ), 0.14 (+), 0.7 ( Q ), and 7 mM ( □) (see Materials and Methods), dialyzed against 10 mM Tris·HCI, pH 7.4/0.1 mM EGTA, and diluted toan A,..,of 1.0. Aliquots were adjusted to the desired magnesium acetate concen· tration and centrifuged at 0° for 10 min at 10,000 rpm in an SS-34 rotor (Sorvall RC-2B). The A260 of the supernatant was measured, and the values are expressed as percentage of total A260. The lines represent nonlinear least-squares exponential fita to the data ( 13). Thermal Denaluration. Acetylated and control chromatins were dialyzed exhaustively against 0.2.5 mM EDTA (pH 8) and adjusted to an A 260 of approximately 1.5. Melting was performed with a Gilford 2400 spectrophotometer equipped with a thermal cuvette and digital absorbance meter under control of a Tektronix-31 programmable calculator. Absorbances were recorded every 0.4 °, each point being the average of 10 readings. The data were corrected for the small hyperchromicity of the buffer and derivatized by using the least-squares method of Savitzky and Golay (11 ) (15-point cubic quartic polynomial first-derivative smooth). The derivative data were normalized to percentage hyperchromicity and resolved into gaussian components by using a nonlinear least-squares fitting program (12, 13). Sucrose Density Gradient Centrifugation of Nucleosomes. lsokinetic sucrose gradients were prepared by the method of Noll (14), assuming a particle density of 1.44 g/ cm3 and using the following parameters: gradient volume, 11.6 ml; C,op, 5% (wt/wt) sucrose; C,.,, 26.7% (wt/ wt ) sucrose; sample volume, 0.5 ml. The gradient buffer was 10 mM Tris-HCI, pH 8/ 0. I mM EGTA. The gradients were run at 36,000 rpm in an SW41 rotor for 18 hr at 4°. The gradients were scanned by using an ISCO UA-5 ultraviolet monitor equipped with a digital absorbance meter, and data were collected by a Tektronix-31 programmable calculator. Fractions of the gradients containing nucleosomes acetylated with [3 H Jacetic anhydride were collected and assayed for radioactivity directly in Aquasol 2 scintillation fluid (New England Nuclear). Polyacrylamide Gel Eleclrophoresb. The products of DNase I digestion of control and acetylated chromatins were FIG. 2. Nuclease sensitivity of acetylated chromatin. Chromatin was prepared from ["C]thymidine-labeled cells and acetylated as indicated in Materials and Methods. The samples were dialyzed against 10 mM Tris,HCI, pH 7.4 , the A26() was adjusted to 0.4, and the sample was adjusted to 10 mM NaCl, 3 mM Mg(OAch, and I mM CaCI,. DNa.oe I at 0.5 µg/ml (l eft) or staphylococcal nuclease at 0.14 µg/ ml (Right) was added , and the digestions were carried out at 24 • . Aliquots (0.5 ml) were removed at intervals and precipitated with I ml of cold 0.3 M perchloric acid. Insoluble material was removed by centrifugation. The supernatant was neutralized with NaOH, adjusted to I% sodium dodecyl sulfate, 50 mM Tris-HCI (pH 7.4), and assayed for radioactivity in Aquasol 2. After I hr of digestion, the remaining chromatin waa adjusted to I% sodium dodecyl sulfate, 50 mM Tris• HCI (pH 7.4), 0.2 M sodium perchlorate and asaayed for total radioactivity. Acetic anhydride concentrations: 0 ( ♦ ), 0.14 mM ( +) , 0.7 mM ( 0 ), and 7 mM (C). The lines represent nonlinear least-squares exponential fitB to the data ( 13) , with the uppermost line in each panel representing the digestion of deproteinized rat liver DNA (see Table I). analyzed on 12% polyacrylamide/7 M urea gels as described by Maniatis et al. (15). Slab gels, 2 X 120 X 150 mm, were prepared and aged overnight. Samples were brought to 50% formamide, boiled for 3 min, and quenched on ice. Electrophoresis was at 200 V and continued until the bromophenol blue dye was 0.5 cm from the end of the gel. The gel was stained for 2 hr with ethidium bromide (I µg / ml ,n 0.5 M ammonium acetate) and photographed under ultraviolet light with a yellow filter. RESULTS One of the most obvious physical changes that accompanies chemical acetylation of rat liver chromatin is its solubilization. Under the conditions of the acetylation reaction (i.e., 0.15 M NaCl), chromatin is a condensed, insoluble precipitate. Chromatin is similarly condensed in the presence of 2 mM Mg 2 +. Upon the addition of acetic anhydride at 0.7 or 7 mM , the chromatin precipitate becomes noticeably more soluble. We investigated this phenomenon by comparing the solubility of control and acetylated nucleosomes in various concentrations of magnesium (Fig. 1). The control nucleosomes were very insoluble, less than 10% remaining soluble in 10 mM magnesium. Acetylation increased the solubility of nucleosomes, the 0.7 mM and 7 mM acetylated material becoming almost com• pletely soluble. Template acti..e regions of chromatin are more sensitive to DNase I (6, 7) than are inactive regions. If acetylation is responsible for gene activation In vivo, In vitro acetylated chromatin might be expected to show greater nuclease sensitivity than unacetylated chromatin. Fig. 2 left and Table 1 show the kinetics of digestion of acetylated 14C-labeled chromatin with DNase I. Increased acetylation of chromatin dramatically al- 16 3246 Biochemistry: Wallace et al. Table l. Proc. Natl. Acad. Sci. USA 74 ( 1977) Kineti c parameters from Fig. 2 DNase I Staphylococcal nuclease k R M k' M' RI Control 0.079 Acetylated chromatin: 0.14 mM 0.237 0.7mM 0.458 7mM 1.:100 DNA 1.290 32.0 1.0 0.054 40.5 1.0 41.6 8:l.3 95.6 100.0 3.9 0.054 15.2 0.110 49.4 0.322 51.3 0.653 53.9 78.9 98.8 100.0 1.3 3.9 14.4 29.6 - "'First-order rate constan t (min - •) . ' Maximal percentage digested. t Relative initial rate of reaction. The initial rate for control chromatin was O.QI A260 unit/ml per min with DNase I and 0.0088 A26() unit/ml per min with staphylococcal nuclease. tered the sensitivity of the DNA to the nuclease. Chromatin acetylated with 0.14 mM acetic anhydride was digested 4 times faster than control chromatin, and that acetylated at 0. 7 mM was digested 15 times faster. Maximally acetylated chromatin (7 mM) was as sensitive as deproteinized DNA to DNase I. Fig. 2 right shows the results of a similar experiment with staphylococcal nuclease. It can be seen that 0.14 mM acetylated chromatin was slightly more sensitive than control chromatin to staphylococcal nuclease, and 0. 7 mM acetylated chromatin was digested 4 times as fast as control chromatin; 7 mM acetylated chromatin was digested approximately half as fast as deproteinized DNA. Thus, chemically acetylated chromatin is moderately sensitive to staphylococcal nuclease but extremely sensitive to DNase I. Weintraub and Groudine (6) reported that, during the digestion of active genes with DNase I, the digested DNA appears as multiples of 10 nucleotides. Fig. 3 shows that 0.14 mM and 0. 7 mM acetylated chromatins were digested by DNase I to the same IO-nucleotide repeat pattern as control chromatin. This result suggests that the histone-DNA interaction that produces this periodic pattern is not disrupted by the acetylation. On the other hand, the 7 mM acetylated chromatin did not have this repeat pattern (although faint bands are apparent at IO-nucleotide intervals). The DNA of the transcriptionally active regions of chromatin Table 2. Summary of melting transitions from Fig. 4 Tm I 'lb Tm2 Control Acetylated: 0.14mM 0.7mM 7mM DNA 64.5 Nucleosomes 34.4 76.3 60.6 81.1 Control Acetylated: 0.14 mM 0.7mM 7mM DNA 62.0 60.2 39.3 73.8 48.3 48.0 65.7 42.1 55.2 52.2 40.8 100.0 'lb Tm3 52.0 79.7 39.0 75.7 31.6 69.7 Chromatin 29.1 71 .1 18.7 77.1 53.3 20.8 65.0 47.6 44.9 63.3 43.7 83.7 50.7 40.8 100.0 26.4 73.6 28.7 6.4 'lb Tm4 'lb 5.0 8.8 13.0 13.2 23.4 83.4 28.8 20.5 81.3 79.8 75.6 :12.3 26.4 9.9 Temperatures are in degrees Celsius. Transitions are numbered in order of increasing temperature. Percentages refer to the fraction of the total hyperchromicity of each sample in a given transition. FIG. 3. Polyacrylamide gel electrophoresis of DNase I-digested chromatin DNA. Control and acetylated chromatins were prepared as described in Material, and Methods. The samples were dialyzed against 10 mM Tris-HCI (pH 7.4) and adjusted to A= of 10. Each was brought to 10 mM NaCl, 3 mM Mg(OAc)2, and 1 mM CaCl2, and DNase I, 10 1<g/ml, was added. Each sample was digested for various times at 24° such that the percentage of acid-soluble A= was 25, 30, 15, and 15 for control, 0.14 mM, 0.7 mM, and 7 mM acetic anhydride, respectively. Reactions were stopped by bringing each to 1% sodium dodecyl sulfate; then the samples were phenol/chloroform extracted as described by Marmur (29). DNA was precipitated with ethanol and dissolved in gel buffer (15); 25 1'8 of each was brought to 50% form amide. boiled for 3 min, and coo led on ice. Electrophoresis was per• formed as described in Materials and Methods. From left to right: DNA from unacetylated chromatin and 0.14 mM, 0.7 mM, and 7 mM acetylated chromatin. is believed to have a lower melting temperature than the DNA of the inactive regions. This belief is supported by the thermal fractionation experiments of McConaughy and McCarthy (16). In addition, DNase II-fractionated template-active chromatin melts at a lower temperature than does unfractionated chromatin (ref. 17; unpublished data). The thermal denaturation of control and acetylated chromatins and nucleosomes, as well as deproteinized DNA, is shown in Fig. 4. The data are presented as first-derivative melting profiles, fitted to three or four gaussian components to quantitate the transitions observed. Table 2 summarizes the melting transitions and the contribution of each to the total hyperchromicity. The most obvious effect of chemical acetylation on thermal denaturation is the dramatic shift to lower melting temperahrres of all the observed transitions. This is true for both chromatin and nucleosomes, although the melting profiles of control an_d acetylated chromatin were more complex. Fig. 5 shows the effect of acetylation on the sedimentation behavior of nucleosomes. The sedimentation of acetylated 17 Biochemistry: Wallace et al. Proc. Natl. Acad. Sa. USA 74 (1977) 3247 LW I A ~::: LJ·-~1500 1 0 i~4500;: ;;; " • ~ 1500 ~ 0 D ~ 4500 D <: 3000 1500 0 a 30 40 50 60 70 80 90 30 40 Temperature, °C 50 60 70 80 90 FIG. 4. First-derivative melting profiles (see Materials and Method$) . Full scale for each profile is 3% change in hyperchromicity per degree. From top to bottom: unacetylated material, 0.14 mM, 0.7 mM, and 7 mM acetylated material, and deproteinized rat liver DNA. (Left) Nucleosomes. (Right) Chromatin. • • •, dH /dT. -, Gaussian components. nucleosomes was dramatically retarded compared with that of control nucleosomes and approached that of sodium dodecyl sulfate-treated control nucleosomes. To determine whether all of these nucleosomes were acetylated equally and whether all of the acetate remained bound to the nucleohistone, the acetylation reaction was carried out with i3Hjaoetic anhydride. Fig. 5 also shows the distribution of 13 Hlacetate across the gradients. From 0.14 to 0.7 mM acetic anhydride there was a 4-fold increase in acetate incorporation per nucleosome whereas from 0. 7 to 7 mM acetic anhydride there was only a 2-fold increase. This demonstrates the limit acetylation reported by Wong and Marushige (10). All nucleosomes, monomer through trimer, were acetylated. At 7 mM acetic anhydride, 13 H !acetate counts were found at the top of the gradient, perhaps representing dissociated histone HI. DISCUSSION We have demonstrated that chemical acetylation of nucleohistone In vitro results in dramatic changes in its physical properties. The results obtained suggest possible functions for acetylation in vivo. The chemical acetylation of nucleosomes was found to increase their solubility in the presence of Mg2+ . Previous studies from this laboratory have shown that DNase II digestion followed by precipitation with 0.15 M NaCl ( 18) or 2 mM MgCl2 (19) yields a fraction of rat liver chromatin enriched in sequences transcribed In vivo (19). 0.5 1.0 Fraction of gradient FIG. 5. Sucrose gradient sedimentation of nucleosomes. Sedimentation is from left to right. From top to bottom: nucleosomes, 0.14 mM acetylated nucleosomes, 0.7 mM acetylated nucleosomes, 7 mM acetylated nucleosomes, and sodium dodecyl sulfate-treated nu cleosomal DNA (30). The sedimentation coefficients of the monomer peak are 11.3, 10.6, 8.6, 6.2, and 5.4 S, respectively. The specific activity of the labeled samples was 19.1, 67.4, and 119 acetates per 200 base pain of DNA for 0.14 mM, 0.7 mM, and 7 mM acetylated nucleosomes, respectively. Under the conditions of acetylation, approximately half as much acetate was incorporated into nonhistones as into histones, and no detectable acetate was incorporated into DNA . The sensitivity of chromatin to DNase I and, to a lesser extent, to staphylococcal nuclease is significantly increased after treatment with acetic anhydride. Similar nuclease sensitivity properties have been reported for transcriptionally active sequences (6, 7, 20), although the biochemical basis for this phenomenon has not been determined. Marushige (8) reported that the derepression of template activity by chemical acetylation with 0.7 mM acetic anhydride was a result of the acetylation of histones H2A and H2B, with H3 and H4 not contributing significantly. H3 and H4 have been implicated in the protection of multimers of 10 nucleotides of DNA from digestion by DNase I (21). Fig. 3 demonstrates that the IO-nucleotide repeat pattern is observed in 0.7 mM acetylated chromatin even though the rate of digestion by DNase I is dramatically accelerated (Fig. 2 left). This suggests that the interaction of H3 and H4 with the DNA has not been greatly perturbed, whereas the template activity measurements by Marushige suggest that the interaction of H2A and H2B with DNA has been functionally altered by acetylation. Higher levels of acetylation result in a disruption of the IO-nucleotide repeat pattern (Fig. 3). Thermal denaturation of nucleohistone demonstrates the 18 3248 Biochemistry: Wallace et al. dramatic stability of the histone-DNA complex. In order to make DNA sequences available for transcription (or replication), it might be necessary to decrease the stability of this complex (22). Although no direct correlation of nucleohistone thermal denaturation behavior with functional state has been demonstrated, isolated template-active chromatin has a lower melting temperature than does the bulk of chromatin ( 17), and active sequences of chromatin elute from hydroxylapatite at lower temperatures (16). The results presented in Fig. 4 and Table 2 demonstrate that chemical acetylation lowers the melting temperature of nucleohistone, as might be expected if acetylation were correlated with transcriptional activation. The transitions observed in the first-derivative melting profiles are similar to those reported by others (23-25). We have resolved the transitions into gaussian components in order to quantitate the contribution of each transition to the total hyperchromicity. From previous work (2, 23), as well as unpublished observations in our laboratory, tentative assignments may be made to the transitions of control nucleosomes. The two transitions at the highest temperatures are due to the denaturation of the histone octamer- DNA complex. The lowest transition is due to a more heterogeneous set of interactions as ri,flected by the breadth of this transition. One of these contributing interactions is probably HI-ONA complexes (26). Acetylation of nucleosomes with 0.14 mM acetic anhydride had little effect on the profile. The relative contribution of each transition remained unchanged, and the transitions were shifted to lower temperatures, Tl being shifted by the largest amount. This is consistent with the role of HI in the Tl transition, in view of the extensive acetylation of this histone by acetic anhydride (10). The acetylation of nucleosomes with 0.7 mM acetic anhydride resulted in dramatic changes in the melting profile: all three transitions were shifted to lower temperatures and the relative contributions of Tl and T2 were increased at the expense of T3. This may reflect the differential effect of acetylation of H2A/ H2B and H3/ H4 as discussed above. Acetylation with 7 mM acetic anhydride virtually abolished the stabilization of DNA melting by histones, even though histones remain associated with the ONA (refs. 8 and 10; Fig. 5). The thermal denaturation profiles of chromatin show the same general features as those of nucleosomes. At the lowest level of acetylation used (0. 14 mM acetic anhydride), there were dramatic changes in magnesium solubility (Fig. I ) and DNase I sensitivity (Fig. 2 left ). However, staphylococcal nuclease sensitivity (Fig. 2 right), DNase I digestion pattern (Fig. 3), and thermal denaturation (Fig. 4 and Table 2) were only slightly perturbed by the acetylation . Although the sites of chemical acetylation have not been determined, it is possible that this low level of acetylation destabilizes interactions that are qualitatively different from those affected by higher levels. The demonstration of a direct role of acetylation of histones in gene activation, as suggested by Allfrey et al. (3), has been elusive. Jackson et al. (27) argued that histone acetylation is too extensive to be a specific mechanism of gene regulation. However, histone modification may be important in DNA replication and histone deposition, as well as in gene activation (4, 28). The many functions of acetylation might be accomplished through different classes of sites for acetylation on the histones. For gene activat ion, it may be sufficient to alter histone-histone or histone- DNA interactions slightly as in the 0.14 Proc . Natl. Acad. Sci. USA 74 (1977) mM acetylation. This could result in the conversion of nucleosomes to a conformation more conducive to transcription of the associated DNA. On the other hand, DNA replication or histone deposition might require a more extensive destabilization of these histone-DNA interactions, as in the higher levels of acetylation reported here. We thank Dr. S. K. Dube for many stimulating di.,cussions and crit ical evaluation of this work and Ms. M. J. Johnson for valuable technical assistance. This work was supported by U.S. Public Health Service Grant GM-13762. R.B.W. is the holder of a Medic,.) Research Council of Canada f,llowship. T D.S. is StJpported by a predoctoral fellowship from the N•lio...J Sc.-ienu, Foundation. I. 2. 3. 4. 5. 6. 7. 8. 9. 10. 1 I. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. 25. 26. 27. 28. 29. 30. Huang, R. C. C. , Bonner, J. & Murray, K. (1964) ). Mot. Biol. 8, 54- 64. Shih, T. Y. & Bonner, J. (1970)). Mot. Biol. 48, 469-487. Alllrey, V. G., Faullmer, R. & Mirsky, A. E. (1964) Proc. Natl. Acad. Sd. USA 51, i86-794. Ruiz-Carrillo, A., Wangh, L. J. & Allfrey, V. G. (1 975) Science 190, 117-128. Weintraub, H. (1975) Proc. Natl. Acad. Sci. US A 72, 12121216. Weintraub, H. & Groudine, M. (1976) Science 193, 84S--856. Carel, A. & Axel, R. (1976) Proc. Na tl. Acad. Sci . USA 73, 3966--3970. Marushige, K. (1976) Proc. Natl. Acad. Sci . USA 73, 39373941. Plagemann, P. G. W. (1969) Blochtm. Btophy,. Acta 182, 46-56. Wong, T. K. & Marushige, K. (1976) Blochemutry IS, 20412046. Savitzky, A. & Golay, M. J. E. (1964) Ana{ Chem. 36, 16271639. Marquardt, D. W. (1963) ). Soc . Ind. Appl. Math. 11, 431441. Pearson, W., Davidson, E. H. & Britten, R. J. (1977) Nuclelc Acid, Res ., in press. Noll, H. (1967) Nature 215, 360-363. Maniatis, T., Jeffrey, A. & van deSande, H. (1975) Blochemutry 14, 3787-3794. McConaughy, B L. & McCarthy, B. J. (1972) Blochemutry 11, 99S--1003. Duerksen, J. D. & McCarthy, B. J. (1971) Btochemtstrv 10. 1471- 1478. Marushige, K. & Bonner, J. (197 1) Proc. Natl. Acad. Sd. USA 68, 2941-2944. Gottesfeld, J. M., Garrard, W. T., Bagi, G., Wilson, R. F. & Bonner, J. (1 974) Proc. Na tl. Acad. Sci . USA 71 , 2193-2197. Reeves, R. (1976) Science 194, 529-532. Sollner-Webb, B., Camerini-Otero. R. F. & Fels.nfeld. G. (1976) Cell 9, 179- 193. Alberts, B. & Frey, L. (1970) Nature 227, 131 3- 13 18. Li, H-J. & Ronner, J. (1971) Biochemistry 10, 1461-1470. Woodcock, C. L. F & Frado, L-L. Y. (1975) Bwchem. Btophy, . Re,. Commun . 66, 403- 410. Li, H-J., Change, C. & Weiskopf, M. (1973) Blochemutry 12, 1763-1771. Olins, D. E. (1969) J. Mol. Biol. 43, 439-460. Jackson, V., Shires, A.. Chalkley, R. & Granner, D. K. (1975)). Biol. Chem . 1!.541, 48S6--486.1. Louie. A. J., Candido, E. P. M. & Dixon. G. H. (1973) Cold Spring Harbor Symp. Quant. Biol. 38, Bro-819. Marrnur, J. (1961) ). Mol. Btol. 3, 208-218. Smith, C. A., Jordan, J. M. & Vinograd, J. (1971) J. Mot. Btol. 59, 255-272. 19 PART IV Chapter 4. Altered nucleosome spacing in newly replicated chromatin from Friend leukemia cells Chapter 5. Properties of newly replicated chromatin: Analysis using shallow metrizamide gradients 20 Reprinted from Proc. Natl. Acad. Sct. USA Vol. 75, No. 12, pp. 5903-5907, December 1978 Biochemistry Altered nucleosome spacing in newly replicated chromatin from Friend leukemia cells (nucleosomes / chromatin replication/ staphylococcal nuclease) ROBERT F. MURPHY, R. BRUCE WALLACE•, AND JAMES BoNNER Division ol Biology. California Institute of Technology, Pasadena, California 91125 Contributed by Jame, Bonner, September 18, 1978 ABSI'RACT Chromatin from Friend leukemia cells labeled with [1'C)thymidine for 24 hr followed by [3Hlthymidine for 10 min is converted into nucleosomes by staphyfococcal nuclease at only half the rate that total chromatin is converted. Polyacrylamide gel electrophoresis of nucleosomal DNA from cells labeled for 24 hr with [14C]thymidine followed by 10 min with [ 3H]thymidine demonstrates that the intemucleosomal spacer of newly replicated chromatin is approximately 20 base fain shorter than that of total chromatin. The implications o this difference for models of chromatin structure are discussed. vol of ice-cold Tris/NaCl/Mg. The reaction mix was centrifuged at 2000 X g for 2 min, and the soluble material was re-moved (this fraction was found to be equivalent to the fraction soluble in 5% perchloric acid). The pellet was suspended in 2.5 mM Tris•HCI, pH 8.0/2.5 mM ethylene glycol bis(#-aminoethyl ether)-N,N,N',N' -tetraacetic acid (EGTA). The Tris/EGTAinsoluble (undigested) fraction was removed by centrifugation at 2000 X g for 3 min. The amount of DNA in the various fractions was determined by bringing aliquots to 1% sodium dodecyl sulfate and scintillation counting in Aquasol-2 (New England Nuclear). A 14C to 3H spilldown correction of 17% was performed (the labeling conditions produced roughly equal amounts of 3 H and 14C cpm). Under these counting conditions, the sum of the cpm in the three fractions for each isotope was constant from low to high enzyme concentration, and equal to the total input cpm. Thus the differences observed were not due to preferential quenching of large i3H]DNA. In addition, no differences between 24-hr 14C-labeled and 24-hr 3 H-labeled DNA were seen (data not shown): Preparation of DNA and Gel Electrophoresis. DNA was extracted from digested nuclei essentially by the procedure of Marmur (20) and dissolved in 89 mM Tris/ 89 mM boric acid/ 2.5 mM EDTA containing 5% (vol/vol) glycerol and 0.01% bromophenol blue. Approximately 10 µg of the DNA samples was applied to 15-cm 4% polyacrylamide gels fin Tris/ borate/EDT A, (21)1 and electrophoresed at 150 V for 2 hr. Gels were fractionated into 2-mm slices (Aliquogel fractionator, Gilson) and scintillation counted in Aquasol-2. Data were corrected for spilldown of 14C and converted to mobility relative to the dye. The replicative form 2 DNA of phage ,/>Xl74 was cleaved with restriction endonuclease Hae Ill and labeled by a slight modification of the procedure of Berkner and Folk (22). Considerable evidence from a number of laboratories indicates that the bulk of eukaryotic DNA exists in a repeated, globular form [reviewed by Kornberg (l)]. The repeat unit (the nucleosome), contains 140 base pairs of DNA associated with an octamer of the "inner" histones (H2A, H2B, H3, and H4), and between 40 and 80 base pairs in an internucleosomal spacer. Recent experiments suggest that 30-50 base pairs of this spacer are complexed with histone Hl (2-4). Variations in the spacer length have been observed between species and cell types (S-11), and the breadth of the bands observed in DNA gels has led to the postulation of heterogeneity within a single cell type (12). Investigations have so far revealed no differences in nucleosome spacing between transcribed and untranscribed chromatin (13-18). In this report we present evidence of a difference in repeat length between newly replicated and total chromatin from Friend leukemia cells. MATERIALS AND METHODS Cell Culture and Labeling. Friend leukemia cells, clone FSD-3, were grown in suspension culture as described previously (19). Cells were labeled with [14C]thymidine (61 Ci/mol, Moravek Biochemicals) at 0.1 µCi/ml for 24 hr (1.5 generations) followed by [3 H]thymidine (21 Ci/mmol, Amersham) at 10 µCi/ml for IO minor at 4 µCi/ml for30min (1 Ci= 3.7 X 10' 0 becquerels). Preparation of Nuclei and Nucleosomes. Cells were harvested at the end of the 3 H labeling period, washed once in phosphate-buffered saline, and suspended in IO mM Tris•HCI, pH 8.0/ 10 mM NaCl/5 mM Mg(OAch/ 0.5% Nonidet-P40 (Shell Chemical). After IO min, a nuclear pellet was formed by centrifugation at 2000 X g for 2 min. The nuclei were washed once in Tris/NaCl/Mg/Nonidet-P-40 and once in Tris/ NaCl/Mg and were resuspended in Tris/NaCI/Mg/0.25 mM CaCl2 . The suspension was adjusted to approximately 50 µg of DNA per ml and brought to 37°C. Staphylococcal nuclease (P-L Biochemicals) was added to aliquots of the nuclear suspension at concentrations of 0.01-6.5 units/ ml (1 unit produces 15 Aiw1 units of acid-soluble material from DNA in 30 min at pH 8.8, 37°C), and digestion was stopped after 5 min by addition of 4 RESULTS We have examined the digestion by staphylococcal nuclease of DNA in isolated nuclei from cells that had been labeled for 24 hr with [1"C]thymidine followed by 10 or 30 min with l3H]thymidine. The amount of radioactivity in acid-soluble, Tris/ EGTA-soluble and Tris/ EGTA-insoluble form was determined at various ratios of enzyme to substrate. [From analysis of sucrose gradient profiles, the Tris/EGTA-soluble fraction represents nucleosome multimers from I to about 20 (data not shown).] Fig. l and Table l show that 30-min-labeled chromatin is digested into nucleosomes at about% the rate, and 10-minlabeled chromatin at less than ½ the rate that total chromatin is digested. However, the fact that there is no difference in the production of acid-soluble material suggests a difference in the The publication costs of this article were defrayed in part by page charge payment. This article must therefore be hereby marked "adverUsement" in accordance with 18 U. S. C. ~1734 solely to indicate this fact. Abbreviation: EGTA, ethylene glycol bis(,'>-amiooetbyl ether)N.N,N' ,N' -tetraacetic acid. • Present address: Department of Biology, City of Hope National Medical Center, Duarte. CA 91010. 5903 21 5904 100 Biochemistry: Murphy el al. Proc. Natl. Acad. Set. USA 75 (1978) A • • • E fr § 100 B ti- • 2.5 log enzyme cone. (units /m l) 2.5 log enzyme cone. !units/mt) FIG. I. Digestion of newly labeled and total chromatin by staphylococcal.nuclease. The percent of total cpm in each fraction (P) is plotted versus the log of enzyme concentration (E, units/ml). (This semilogarithmic plot expands the low enzyme region.) (A) Ten-minute labeling, 3 H total cpm • 35,421 (±6.3%) , 14C total cpm • 13,7 10 (±8.3%) (uncertainties are 'lb SD). (B) Thirty-minute labeling, data from three experiments. 3 H total cpm • 164,524, I 10,084, 167,690; "C total cpm • 46,223, 100,724, 100,750 (all ±3-4%). ♦, 14C (total) Tris/ NaCl/Mg-soluble (acid-soluble); O, 3 H (new) Tris/NaCl/Mgsoluble (acid-soluble) ; ■ , 14C (total) Tris/EGTA-soluble (nucleosomes); □, 3 H (new) Tris/EGTA-soluble (nucle<JSOmes) . The lines represent fits to the equation P - Me•E drawn hy using a nonlinear least -squares fitting program (R. F. Murphy, W. R. Pearson, and J. Bonner, unpublished and ref. 23) (see Table 1). 1 initial availability of intemucleosomal DNA. The accuracy of these estimates depends in large part on the adherence of nuclease digestion to pseudo-first-order kinetics under our conditions. This problem may be avoided by plotting the same data as 14C/ 3H ratio (Fig. 2). It can be seen that the difference is Table I. Kinetic parameters from Fig. 1 k• M' RI 10-min total Ratio new/total 0.224 0.416 0.539 70.2 76.9 0.912 15.7 32.0 0.491 30-min new 30-min total Ra.tio new/total 0.665 0.956 0.696 61.4 66.8 0.920 0.640 Chrbmatin 10-min new • First-order rate constant (ml unit- • min- 1). t Maximum percent digested. I Initial rate of reaction (/c X M ). 40.9 6:\.9 FIG. 2. Comparison of digestion of total and newly la beled chromatin. Data from Fig. I replotted as "Cf'H ratio for each sample. ♦ , 10-min Tris/NaCl/Mg-soluble; O, 30- min Tris/NaCl/Mg-soluble; ■ , 10-min Tris/EGTA -soluble; □, 30-min Tris/EGTA -soluble. greatest at low enzyme-to-chromatin ratios, approaching 8-fold for a 10-min labeling. The difference in digesiion rates observed suggests a difference in the structure of newly replicated chromatin. In order to examine that possibility, total DNA from various digests of nuclei labeled for 24 hr with l1"C]thymidine followed by 10 min with [3H]thymidine was prepared and electrophoresed in 4% polyacrylam!de gels (Fig. 3). The monomer peaks in each gel are coincident, but a difference in the higher multimers is evident. The 3H peaks appear broader, so that for dimers and trimers the 1•c peak is almost included in the 3H peak. The panels on the right show the data plotted as 3H/ 1"C ratio. Peaks are seen to the right of the position of the 1"C-labeled multimers (marked by the arrows), demons.trating a difference in the average size of the two types of multimer DNA. To quantitate this difference, Gaussian curves were fit to the data from each gel for each isotope by using a nonlinear leastsquares fitting program (R. F. Murphy, W. R. Pearson, and J. Bonner, unpublished; ref. 24). 32 P-Labeled fragments of ,t,Xl74 replicative form 2 DNA cleaved with Hae III were either run on parallel gels or mixed with the 3 H and 14C samples. These gels were also fractionated and their radioactivities were determined. The mobilities of these fragments were 1- 2% higher at low DNA concentrations (approximately 100 ng/ gel), compared to the same fragments mixed with the Friend cell DNA samples (approximately 10 µg / gel; data not shown). For the standardization, values from the latter gels were used. Fig. 4 shows the nucleotide length of each fragment Idetermined from the nucleotide sequence (25)] plotted semilogarithmically versus its mobility. This yields a straight line, from which the sizes of the 3 H and 1 ◄c nucleosomes were calculated. Fig. 5 shows the variatioo in uucleosome size with multimer number, from which the nucleowme repeat length can be de- 22 Biochemistry: Murphy el al. 15 A Proc. Natl. Acad. Sci. USA 75 (1978) 560 360 160 140 .j, .j, H D 560 360 .j, .j, 5905 160 140 H 2 19 0 .5 E E -~ " u ;;; ~ §. * 2 f j 0 .5 28 F 2 0 .5 0 .5 0 Relative mobility FIG. 3. Polyacrylamide (4%) gel electrophoreais of DNA from newly labeled and total nucleosomes. (A -C) Percent of total nuclear cpm for 'C (0, total) nucleosomal DNA is plotted versus mobility relative to bromophenol blue. 3 H total cpm: A, 4019; B , 4571; C, 3605. " C total cpm: A, 5772; B, 6259; C, 4873. The lines represent the sum of Gaussian curves fit to the data by using a nonlinear least-squares fitting program (R. F. Murphy, W. R. Pearson, and J . Bonner, unpublished). T he arrows mark the approximate position of the "C core and multimers, and the associated numbers of base pairs were calculated from the standards as described in the text. The enzyme concentration increases 3.16-fold from A to Band from B to C. (D - F) Ratio of3H/ "C cpm from A-C. 3 H (+ , new) and 1 termined independently of the extent of digestion. Table 2 summarizes this data. The difference in size is clearly due to a difference in spacer length. which varies from 10 base pairs in the least digested sample to 29 in the most. Because the breadth of the bands appears to decrease with digestion, the average value of 17 is probably an underestimate. Table 2. Al Rt Ct Nucleosomal DNA sizes Chromatin M ono Di Tri Spacer Unit Spacer• difference New Total 168 164 387 397 563 579 198 208 58 68 New Total New Total 151 150 144 148 349 365 331 358 528 548 188 199 187 210 48 59 47 70 New Total New Total 140 139 138 140 323 348 320 . 336 527 585 194 223 182 195 54 83 42 55 Average 10 11 23 29 13 17 ± 8 • Assuming a core size ol 140 base pain. t Fig. 3 panel number. I Gel not shown; enzyme concentration as for preceding gel. DISCUSSION Much evidence has been accumulated to indicate that nucleosomes from different species and cell types have different repeat lengths (5--11). A correlation between nucleosome size and transcriptional activity has been suggested (9- 10). As has been pointed out by Thomas and Thompson (11), however, it would appear unlikely that two different cells from the same organism would have a significant difference in their average repeat size because of differences in the small fraction of total DNA coding for structural genes. Some reports of the presence of transcribed genes in nucleosome structures have included data that indicate that the spacing for transcribed and nontranscribed DNA is the same. Two of these methods-com· parison of unlabeled DNA in parallel gels (17) and comparison of unlabeled total DNA and labeled, hybridized probe for the examined sequence (15, 16, 18)-are severely limited in their accuracy due to difficulties-in aligning and scaling the different sets of data. The third method-double-labeling- has been used to show that 60-to 120-min labeled ribosomal DNA and long-term labeled total DNA from Tetrahymena macronuclei nucleosomes have the same size (13, 14). No direct evidence of a relationship between nucleosome spacing and transcriptional activity has been reported. As we have shown above, newly replicated nucleosomal DNA from mouse Friend cells is different in size from total nucleo-- 23 5906 Biochemistry: Murphy et al. Proc. Natl. A.cod. Sci. USA 75 (1978) 5000.-------------, A B C 500!:-------------J Relative mobility FIG. 4. Molecular weight standardization curve for 4'11> polyacrylamide gels. DNA length L (base pairs) is plotted semilogarithmically versus mobility relative to bromophenol blue for <i>Xl74 replicative form DNA cut with Hae III ( ♦ ,lengths determined from the DNA sequence (24)1. The line represents the least-squares lit log L = 3.308 - 1.473 RF (correlation coefficient r • -0.9963, 'II, e,-ror ~ 1.06) forthe .<i>X 174 markers, whose lengths are 118,194,234,271 and 278,310,606,872, and 1078 base pairs. FIG. 6. Changes in nucleosome arrangement during chromatin replication. All circles represent 140-base-pair nucleosome cores. 0, nucleooomes with normal (200-base-pair) spacing; • • nucleosomes with shortened (180-base-pair) spacing. The number of nucleooomes depicted in each configuration is arbitrary. In reality, the number of nucleooomes (and hence the length of time) between the passage of the replication fork and regaining of normal spacing ii, certainly much larger. (A) Chromatin after replication has been initiated. Compression of nucleooomes has occurred to allow for replication compleL A random dispersive model is shown, but semiconservative and con servative models are not ruled out. Current data are conflicting in this regard, but the propoeed compression during replication is independent of the exact dispersion mechanism. Nuclease digestion of chromatin labeled for short times {1-30 sec, depending on the rate of replication for the specific cell type) yields an increased rate of production of both acid-soluble material and ·mononucleooomes. (B) New nucleooome cores cover the free DNA, but shortened spacing remains intact. Additional free DNA generated by compression is present at the fork. Longer pulses (f>--. 10 min) detect no difference in production of acid-soluble material, decreased oligonucleoeome production, and shortened nucleosome spacing. (C) Normal spacing is slowly regained by nucleosome sliding (20-30 min). The amount of free DNA at the fork is thus kept constant. B -.a. j £ 370 t < z 0 2 3 Nucleosome number FIG. 5. Determination of nucleosome spacing. DNA length (base pairs, determined from the center of fitted Gaussian curves) is plotted versus nucleosome multimer number (n) for the gels from Fig. 3. +, 3 H; o, "C. The lines are lea.st-squares lits L • a + bn, in "'hich a ii, the sum of the lengths of the ends on each multimer minus the spacer length and bis the unit nucleosome size (core length plus the spacer length) (see Table 2). somal DNA. This might suggest a relationship between nucleasome spacing and chromatin replication rate, because the general correlation between shortened nucleosomal spacings and high transcriptional activity [reviewed by Kornberg (1)) can also be made between shortened spacing and short generation times. However, there is no difference in spacing between Chinese hamster ovary cell nuclei and mitotic chromosomes ( 17), or between confluent and exponentially growing C6 rat glial tumor cells (10). These experiments used unlabeled DNA in parallel gels, and hence do not have the sensitivity of our experiments. However, the conclusion that nucleosome spacing is not significantly related to rate of chromatin replication is probably justified, because the fraction of chromatin in newly replicated form is too small to account for spacing differences in total DNA from cells with different generation times. Instead, a gradual increase in nucleosome spacing with time is possible (perhaps to compensate for small amounts of histone degradation or as a result of changes in the level of histone modification). Some laboratories have reported that chromatin pulse-labeled for from 0.5 to 10 min produces acid-soluble material at an increased rate (relative to total chromatin) when digested with staphylococcal nuclease (26-28). These studies have generally used shorter pulses than we have, and those that have used longer pulses have indicated that the difference disappears (26, 24 Biochemistry: Murphy et al. 28). Considering the variation in chromatin replication rate among different cell lines, it would appear that the published data are consistent with an increased sensitivity during and immediately following DNA replication and normal sensitivity once histone cores are deposited on the free DNA. The shortlived acid-soluble difference Is followed by a longer-term decrease in production of nucleosomes by staphylococcal nuclease, due to smaller nucleosome spacers (Fig. 6). In any case, our results indicate that elongation can take place after histone deposition (perhaps caused by the binding of histone Hl and/or some nonhistone protein). This suggests that the interaction between core histones and DNA is not strong enough to prevent nucleosorne "sliding." a conclusion that may be sig.. ificant for models of chromatin replication and transcription. After this manW1Cript was prepared, Levy and Jakob (29) and Seale (30) reported similar results for sea urchin embryos labeled for 7 sec and HeLa cell nuclei labeled m mtro for 20 min, respectively. ,j,Xl74 DNA cleaved with Hu Ill was the gift of Richard Parker. We thank Thomas Sergent for labeling the 4>Xl74 DNA and for helpful disctmion of this manll5Cript. This work was supported by U.S. Public Health Service Grant GM-13762. R. F.M. is a National Institutes of Health predoctoral traiMe. R.B. W. is the holder cl a Medical Research Council of Canada Centennial fellowship. 1. 2. Kornberg. R. D. (1977) Annu. RIIO. Blochem. 46, 931-llM. Whitlock, J. P., Jr. & Simpson, R. T. (1976) Blochemlltry 15, 3307-3314. 3. Bakayev, V. V., Bakayeva, T. G. & Varshavsky, A. J. (1977) Cell 11,61~29. 4. Noll, M. & Kornberg, R. D. (1977) J. Mol. Biol. 109, 393-404. 5. Noll, M. (1976) Cell 8, 3 4 ~. 6. Morris, N. R. (1976) Cell 8, 357~. 7. Thomas, J. 0 . & Furber, V. (1976) FEBS Lett. 66, 274-280. 8. Spadafora, C , Bellard, M., Compton, J. L & Chambon, P. (1976) FEBS Lett. 69, 281- 285. 9. Morris, N. R. (1976) Cell 9, 627-632. Proc. Natl. Acad. Sci. USA 75 (1 978) 5907 10. Compton, J. L, Bellard, M. & Chambon, P. (I 976) Proc. Natl. Acad. Sci. USA 73, 4382- 4386. 11. Thomas, J. 0 . & Thompson, R. J. (1977) Cell 10, 633-640. 12. Lohr, D., Corden, J., Tatchell, K., Kovacic, R. T. & van Holde, K. E. (1977) Proc. Natl. Acad. Sci. USA 74, W-8:l. 13. Piper, P. W., Celis, J., Kaltoft, K., Lerr, J. C., Nielsen, 0 . F. & Westergaard, 0. (1976) Nucleic Ac1<U Re,. 3, 493-50.5. 14. Mathis, D. J. & Gorovslcy, M. A. (1976) Biochemistry 15, 75075.'I. 15. Bokhon1:o, A. & Reeder, R. H. (1976) Blochem . Blophys. Re, . Commun. 70, 14~152. 16. Ba!tock, CJ., Christie, S. & Hatch, F. T. (1976) Nature (London) W,51~19. 17. Compton, J. L. , Hancock, R., Oudet, P. & Chambon, P. (1976) Eur. J. Blochem. 70, 555- 568. 18. Gottesfeld, J. M. & Melton, D. A. (1978) Nature (London) 273, 317--319. 19. Dube, S. K., Gaedicke, G., Kluge, N., Weimann, 8. J., Melderis, H., Steinheider, G., Crozier, T. & Ostertag, W. (1974) in Proceeding, of the 4th Sympo,jum: Princess Takamatsu Cancer Research Fund, Differentiation and Control of Malignancy in Tumor Cell, (Univ. of Tokyo Press, Tokyo), pp. 99-136. 20. Marrnur, J. (1961) J. Mol. Biol. 3, 208-218. 21 . Maniatis, T., Jeffrey, A. & van deSande, H. (1975) Biochemistry 14, 3787 --3794. 22. Berkner, K. L & Folk, W. R. (1977) J. Biol. Chem. 252, 317~ 3184. 23. Pearson, W. R., Davidson, E. H. & Britten, R. J. (19i7) Nucleic Aclda Re,. 4, 1727- 1737. 24. Wallace, R. 8 ., Sargent, T. D., Murphy, R. F. & Bonner, J. ( 1977) Proc. Natl. Acad. Sci. USA 74, 3244--3248. 25. Sanger, F., Air, G. M., Barrel, B. G., Brown, N. L., Coulson, A. R., Fiddes, J. C. , Hutchison, C. A., lll, Slocombe, P. M. & Smith. M. (1977) Nature (London) 265, 687- 695. 26. Seale, R. L. (1975) Nature (London) 255, 247- 249. 27. Hildebrand, CE. & Walters, R. A. (1976) Blochem. Btophr;s. Res. Commun. 73, 157-163. 28. Burgoyne, L. A., Mobbs, J. D. & Marshall, A. J. (1976) Nucleic Ac1dl Rn. 3, 3293--3304. 29. Levy. A. & Jakob, K. M. (1978) Cell 14, 25S-267. 30. Seale, R. L. {1978) Proc. Natl. Acad. Sci. USA 75, 2717- 2721. 25 PROPERTIES OF NEWLY-REPLICATED CHROMATIN: ANALYSIS USING SHALLOW METRIZAMIDE GRADIENTSt Robert F. Murphy*, R. Bruce Wallace* and James Bonner Running title: Properties of newly-replicated chromatin +From the Division of Biology, California Institute of Technology, Pasadena, CA 91125. This work was supported by U.S. Public Health Service Grant GM-13762. +Present address: Division of Biology, City of Hope National Medical Center, 1500 East Duarte Road, Duarte, CA 91010. 26 Abstract The properties of chromatin containing newly synthesized DNA and protein are investigated. Nuclease digestion rates suggest that histones are segregated conservatively at the replication fork. A fraction of soluble nucleosomes enriched in newly-replicated DNA has been isolated by means of its increased density in metrizamide relative to bulk chromatin. This fraction is shown to be packaged into nucleosomes, but at an interval of approximately 160 base-pairs, as opposed to the 200 base-pair repeat seen for total chromatin. Evidence is presented that the density difference is due to this altered spacing. While pulse-labeled DNA is present in this dense fraction, nucleosomes labeled with short pulses of arginine or acetate are of normal density. The data presented are consistent with the conclusion that newly replicated DNA is associated with pre-existing histones in a short-lived, compact structure, while newly-synthesized histones are deposited at normal spacing some distance from the replication fork. 27 The events involved in the replication of eukaryotic chromatin hav e been analyzed by fixation of chromatin with formaldehyde (e.g. , Fakan et al., 1972; Jackson et al., 1975; Seale, 1976a; Hancock, 1977) and inhibition of protein synthesis with cycloheximide (e.g., Seale & Simpson, 1975; Weintraub, 1976). results. Some of the se investigations have yielded conflicting In addition, possible disruption of chromatin structure resulting from the methods of sample preparation and analysis used has led to objections to the results of previous experiments. In this paper we describe the determination of some of the properties of replicating chromatin using metrizamide density gradient centrifugation of native chromatin. 28 Materials and Methods Cell culture and labeling . Friend leukemia cells, uninducible clone F4+, were grown in suspension culture as described previously (Keppel et al., 1977) . under our conditions was 15.2 hr. were labeled with 14 The generation time For various experiments, cells c-thymidine (61 Ci/rmnol, Amersham) at 4 nCi/ml, 14 c-L-arginine (312 mCi/rrnnol, Schwartz/Mann) at 50 nCi/ml, 3 H-thymidine (22 Ci/nnnol, Amersham) at 40 uCi-min/ml, 3 H-L-arginine (8.8 Ci/nnnol, Amersham) at 250 uCi-min/ml, and 3 H-acetic acid (2 Ci/nnnol, Moravek Biochemicals) at 300 uCi-min/ml (1 Ci=3.7 x 10 10 becquerels). All samples for scintillation counting were brought to 0.1% SDS in a final volume of 0.4 ml and counted in 3 ml of Aquasol-2 (New England Nuclear). Under these conditions, the spillover from the 3 H window to the 14 c windo w was 4%, and the 14 c to 3H spillover was 11%. No effect on the counting efficiencies or spillover ratios was seen from 0-50% metrizamide. The presence of the SDS eliminated almost all self-quenching of large molecul es. Preparation of nuclei and nucleosomes . Nucleosomes were prepared from purified nuclei as described previously (Murphy et al., 1978), with the e xception that micrococcal nuclease was purchased from Worthington (1 unit produces 1 A260 unit of 0 acid-soluble material from DNA in 30 min at pH 8.0, 37 C). facilitate comparison of different To experiments, digestions were carried out for 5 min at 37°c at specified ratios of nuclease units to number of nuclei. 29 Metrizamide gradient centrifugation. Metrizamide was obtained from Nyegaard and Co. A/S, Oslo, Norway. All metrizamide solutions contained 2.5 mM Tris pH 8.0/ 2.5 mM ethyleneglycol-bis-( -aminoethyl ether)N,N'-tetraacetic acid (EGTA). 2.3 ml of sample in 30% (w/v) metrizamide was layered over 1.7 ml of 60% (w/v) metrizamide in a~ x 2 in cellulose nitrate centrifuge tube, and overlayered with 1 ml of Tris/EGTA. Gradients were run at 10,000 rpm for 16 hr at 4°c in the Sorvall TV-865 rotor, and fractions collected dropwise from the bottom. Density was determined from the refractive index at 25°c (Birnie, 1978). DNA samples for electrophoresis were prepared by proteinase K (EM Laboratories) digestion of gradient fractions. Polyacrylamide gel electrophoresis and fluorography. 4% polyacrylamide slab gels containing 89 mM Tris/89 mM Borit acid/ 2.5 mM EDTA (pH 8.3) were prepared as described by Maniatis et al. (1975). 3 H radioactivity was visualized by fluorography (Bonner & Laskey, 1974; Laskey & Mills, 1975). 2.5% polyacrylamide/ 0.5% agarose Tris/Borate/EDTA tube gels were prepared similarly. Tube gels (0.6 x 15 cm) were run for 3 hr at 100V and fractionated into 2 mm slices (Aliquogel fractionator, Gilson). Gel fractions were counted as described above. polymerized with various sizes of 3 H and 14 Counting of gels c DNA showed no quenching by the polyacrylamide fragments, and minimal dependence of the efficiency and spillover on DNA length. 30 Results Nuclease sensitivity of pulse-labeled chromatin. In a previous paper (Murphy et al., 1978), we demonstrated that chromatin pulse-labeled for 10 min was digested to nucleosomes more slowly than bulk chromatin, using the criteria of solubility in Tris/EGTA. We attributed this to the shortened nucleosome spacing of newly-replicated chromatin (Murphy et al., 1978; Levy & Jakob, 1978; Seale, 1978b). While an increased rate of digestion to acid-soluble DNA by micrococcal nuclease had been reported for pulse-labeled chromatin (Hildebrand & Walters, 1976; Seale, 1976b), we did not observe this difference for our 10 min labeled material. Because of possible differences in replication rates between different cell lines, we have determined the relative rates of production of acid-soluble and Tris/EGTA soluble material for cells labeled for 1-30 min. Figure 1 shows that digestion of newly-replicated chromatin results in an increased rate of production of acid-soluble material, but a decreased rate of production of soluble nucleosomes. In view of the smaller spacing and decreased nucleosome production of newlyreplicated chromatin, we attempted to fractionate unfixed nucleosomes using metrizamide gradients. Buoyant density of pulse-labeled chromatin. Cells were labeled with 14 c-thymidine for 24 hr, and then with 3 H-thymidine for 1, 10 or 100 min. After brief digestion with micrococcal nuclease, the total nucleosomal fraction was centrifuged in metrizamide. Figure 2 shows that the 1 min 31 B 7 0 0 5 5 0 0 !I,... !I,... l) I 'Q'" ......... , . r<) :c 03 3 r<> 'Q'" -8 -5 -8 -5 log enzyme cone. (units/nucleus) Figure 1. Digestion of pulse-labeled chromatin with micrococcal nucl e ase. Cells were labeled for 24 hr with 14 c-thymidine, followed by 1 min (0) 5 min (D ), 10 min (X), or 30 min(+) with 3 H-thymidine. A) i4c/3H ratios for Tris/N 9Cl/Mg soluble fraction (equivalent to acid soluble). B) 3 H/ 14 c ratios for Tris/EGTA soluble nucleosomes. 32 - r<>E 1.3 0) 1.2 U I.I 0 ' m t--------------------------~ (\J B +- 0 Q.. E C Cl.. 0 0 E ~ 0 C D 0 0.5 fraction from bottom of gradient Figure 2. Separation of newly-replicated chromatin from bulk chromatin in metrizamide gradients. Cells were labeled for 24 hr with l4c-th midine and then for 1 min (B), 10 min (C) or 100 min (D) with H-t9ymidine. After digestion with micrococcal nuclease at 5.33 x 10 units/nucleus and removal of acid-soluble material (2.4% l4c; 8.8, 2.4, 2.4% 3H), the total nucleosomal fraction was centrifuged in metriz amid e. The data in B-D are plotted so that the total area of each panel represents zeven timej the number of cpm in the gradient. + density. ,¢ 1 C cpm. C H cpm. (See Table I.) 3 33 Table I. Density of chromatin fractions in metrizamide gradients of Figure 2a 3H Heavy 14c %Tb %T Light 14c 3H 1 min 1.2034 28.2 1.2070 2.9 1.1712 1.1664 10 min 1. 2106 10.4 1. 2107 6.0 1. 1730 1. 1701 100 min 1.2097 3.6 1. 2093 3.4 1.1678 1.1669 Mean 1.2079 1.2090 4.1 1.1707 1. 16 78 SD 0.0039 0.0019 1.7 0.0026 0.0020 aPeak densities and percentages were determined using a non-linear least-squares fitting program (Murphy et al., 1979) 0 fit two Gaussian curves to the data. Percent of total cpm in gradient. 6 34 labeled nucleosomes are more dense than the long-term labeled nucleosomes, and that this density difference disappears on longer labeling. Table I gives the densities of the two chromatin fractions, and shows that bulk chromatin has a dense shoulder constituting approximately 5% of the total DNA. Birnie et al. (1973) found the density of native mouse DNA to be 1.118 g/cm3 , and the density of purified proteins to vary from 1.24 to 1.29 g/cm3. The density of nuclear DNA (proteinase K treated) in metrizamide is 1.12 g/cm3 , and that of nucleosomal protein (DNAse I treated) is 1.23 g/cm3 (data not shown). Thus, it is likely that the increased density of newly-replicated chromatin is due to an increased protein/DNA ratio, as would be expected due to its decreased nucleosomal spacing. To confirm that the material in the dense peak is present in nucleosomes, the material from that peak was isolated and run on isokinetic sucrose gradients (Noll, 1967). The 3 tt labeled material showed the typical nucleosomal pattern (data not shown). Figure 3 also shows that the DNA of this fraction has nucleosomal spacing. Since there is more newly-replicated DNA in the mononucleosome bands of the light fractions than in the corresponding bands of the denser fractions, the separation may be at least partly due to a decreased rate of production of low multimers of newly-replicated chromatin. Removal of density difference by extensive nuclease digestion. If the density difference were totally due to the presence of newly-replicated DNA in higher nucleosome multimers, 35 Figure 3 (next page). Nucleosome distribution of metrizamide gradient fractions. Nuclei labeled for 3 min with 3 H-thymidine were digested at 2 x 10 6 units/nucleus (18% acid-soluble), and the tot~l nucleosomal fraction was centrifuged in metrizamide. A) H distribution across gradient. B) 13 day fluorogram of a 4% polyacrylamide slab gel .. The outermost slots contain DNA from an aliquot of the nucleosomes before centrifugation. The other slots (from left to right) contain DNA from the fractions marked by diamonds in A. 36 3200 E ·a.. u I r<) 15 30 45 fraction no. • 37 it would disappear if soluble nucleosomes (i.e., low multimers) were isolated and run. That this is not the case is demonstrated by the fact that 1 min labeled soluble nucleosomes are more dense than long-term labeled nucleosomes (Figure 4). However, the density difference disappears as the chromatin is digested to monomers, as would be predicted if the higher density were due to decreased nucleosomal spacing. Figure 4 also suggests that there is a shift in density of bulk nucleosomes as nuclease digestion proceeds. This is further demonstrated by the data from a number of experiments sunnnarized in Table II. Nucleosome spacing of isolated newly-replicated chromatin. The data presented above show that .it is possible to isolate a fraction of nucleosomes which are enriched in newly-replicated DNA. To confirm that indeed the nucleosome spacing of this fraction is significantly smaller than that of bulk chromatin from the same digest, DNA from these fractions was isolated and analyzed on polyacrylamide gels . (Figure 5). molecular weight from relative mobility using Estimating 32 P-labeled PBR322 DNA digested with the restriction endonuclease HinFl, the repeat spacing of the 1 min labeled nucleosomes is about 160 bp, while that of the bulk chromatin is 200 bp. This is smaller than our previous estimate of 180 bp for 10 min labeled nucleosomes (Murphy et al., 1978). As we pointed out, however, this is due to the reversion to normal spacing which occurs in the 10 min labeling period . 38 A E 0. B E C F • (.) ""O Q) N 0 E ~ 0 C • 0 0.5 I 0.5 fraction from bottom of gradient Figur e 4. Effe c t of e xt e nt of digestion on separation of newly-replicated chromatin from bulk chromatin. Tris/EGTA-soluble nucleosomes from cells ~abeled for 24 hr with 14 c-thymidine (◊) followed by 1 min with H-thymidine ( 0) were centrifuged in metrizamide. __Tl:~ __ total area of e_~~_h_ pane~ __ is_ s _i~ time~ -~~-~ _ total nuc6ear cpm. Digest1zns (A-F) were from 1.67 X 10 to 6.67 x 10 units/nucleus. C acid-soluble increases from 3 0.6% to 23%, and H acid-soluble from 4% to 39%. The lines through the points are 5 point quadratic smooths of the data (Savitzky & Golay, 1964). 39 Table II. Buoyant density of nucleosomes as a function of extent of digestiona Enzyme Rangeb NC Thymidine Mean SD -8 to -7.1 5 1.1577 -7 to -6.1 7 -6 to -5.1 5 N Arginine Mean SD 0.0024 3 1.1580 0.0078 1.1517 0.0059 7 1. 154 7 0.0068 1.1363 0.0049 2 1. 1417 0.0007 aPeak densities were determined by fitting a Gaussian curve to gata for 24 hr labeled material. Log of enzyme concentration (units/nucleus); lower and upper limits for grouped data. cNumber of gradients. 40 2.8 A E a. u vu -0 0 6 B E a. u I rt) -0 0 0 relative mobility Figure 5. Nucleosomal DNA spacing of heavy fraction from metrizamide gradients. 2 . 5% polyacrylamide/0.5% agaros gels. Samples were from the gradient shown in Figure 4C. A) 14 C cpm for Tris/EGTA soluble fraction before centrifugation. B) 3H cpm for he avy peak from gradient (fractions 25&26). Mobilit y is r e lative to bromphenol blue. The arrows mark the positions of 32 P-labeled PBR322 DNA cut with HinFl, and correspond to DNA sizes of 1631, 516/506 , 396, 344, 298, and 221/220 base-pairs. 41 Buoyant density of newly-deposited nucleosomal protein. Some investigators have suggested that newly-synthesized chromosomal protein is deposited on unreplicated DNA (Jackson et al., 1976; Seale, 1976a; Hancock, 1977; Hancock, 1978). There has been some uncertainty about this conclusion, since CsCl gradients of formaldehyde-fixed chromatin were used. To determine the degree of association of newly-synthesized histone with newly-replicated DNA using unfixed chromatin, cells were labeled with 14 c-L-arginine for 24 hr, and then with 3 H-L-arginine for 5 min. Tris/EGTA soluble nucleosomes from these cells were run on metrizamide gradients after digestion to varying extents. Figure 6B shows one of these gradients, along with a gradient of 24 hr 14 c-thymidine/ 5 min 3 H-thymidine labeled Tris/EGTA soluble nucleosomes (at the same nuclease • concentration). I n no case were the dense region. Indeed, the 14 3H • • • -arginine counts d etecte d in C and identical, as can be seen in Figure 6. 3 H counts were nearly Similar results were obtained for cells labeled with arginine for 10 or 100 min, and for cells labeled with acetate for 1, 10 or 100 min. 42 A B E a. u -0 Q) N 0 E l,,_ 0 C 0 0.5 I 0.5 fraction from bottom of gradient Figure 6. Comparison of density distribution of thymidine and arginine labeled Tris/EGTA soluble nucleosomes. Digestion was at 1.0 x 10 7 units/nucleus. A) 24 hr 14 c-thymidine (◊; 2 8% acid-soluble)/ 5 min 3 H-thymidine (0; 3.7% acid-soluble). B) 24 hr 14C-L-arginine ( ◊) / 5 min 3 H-L-arginine (0 ) . The total area of each pane 1 is 6. 6 times the total cpm in each gradient. 43 Discussion Nuclease sensitivity and histone segregation mechanism. Chromatin pulse-labeled with thymidine for short periods o f time has been shown to be digested to acid-soluble material more quickly than bulk chromatin by micrococcal (staphylococcal) nuclease (Hildebrand & Walters, 1976; Seale, 1976b; Levy & Jakob, 1978) and DNase I (Seale, 1975; Burgoyne et al., 1976). This increased production of acid-soluble material has also been demonstrated for chromatin from cells which have been exposed to cycloheximide (Seale & Simpson, 1975; Weintraub, 1976). While Seale (1976b) suggested that newly-replicated chromatin was digested to nucleosomes faster than total chromatin, Marshall & Burgoyne (1976) demonstrated a slightly decreased rate of production of nucleosomes for 10 min labeled chromatin using sucrose gradients. We have confirmed this decreased rate of production of nucleosomes by measuring solubility in Tris/EGTA, which is proportional to fragment length (Murphy et al., 1978 and Figure 1). These seemingly contradictory results can best be explained by assuming a conservative mechanism of histone segregation at the replication fork. Such a mechanism has been demonstrated using cycloheximide treated cells (Weintraub, 1976; Riley & Weintraub, 1979) and by histone-histone cross-linking (Leffak et al., 1977). The fact that other investigators (e.g., Jackson et al., 1975) have presented evidence for a random mechanism may be a result of problems involved in chromatin fixation, or 44 difficulties in correcting for the presence of non-histone protein. Metrizamide fractionation of chromatin. Because metrizamide is an inert, non-ionic density gradient material, it has been used to examine chromatin properties (e.g., Birnie et al., 1973). Rickwood et al. (1974) demonstrated the separation of chromatin into two fractions as a function of the extent of mechanical shearing, and took this as an indication of the presence of protein-rich and protein-poor regions. The data of Figure 4 and Table II suggest that the separation is instead due to the decreased density of low nucleosome multimers. This shift may well be due to the removal of non-histone proteins and/or histone Hl during digestion (Noll & Kornberg, 1977). Buoyant density of pulse-labeled chromatin. In CsCl gradients of formaldehyde-fixed chromatin, a decreased density has been reported for pulse-labeled chromatin (Fakan et al., 1972; Seale & Simpson, 1975; Jackson et al., 1976), cycloheximide-treated chromatin (Seale & Simpson, 1975), and in vitro labeled chromatin (Seale, 1978a). This decreased density in CsCl corresponds to the increased density in metrizamide demonstrated above, since protein is less dense than DNA in CsCl, but more dense than DNA in metrizamide. Both increased (Levy et al., 1975) and decreased (Burke & Pearson, 1976) densities in metrizamide have been reported for pulse-labeled chromatin which had been mechanically sheared. As Noll et al. (1975) have pointed out, there is significant disruption of nucleosome 45 structure during mechanical shearing. In addition, the results presented above demonstrate the importance of controlling the extent of shearing during sample preparation. These facts make interpretation of the previous reports difficult . Hancock (1974) has shown that the preparation of nuclei and chromatin by methods very similar to those used in these experiments do not allow exchange of histones between deoxyribonucleoprotein molecules. The lack of histone exchange, the fact that nuclease digestion was used to prepare nucleosome samples, and the use of the non-ionic density gradient material make it very unlikely that the results presented above are due to distortion or disruption of the chromatin structure. Deposition of newly-synthesized histone. Having established a method for isolating native nucleosomes containing newly-replicated DNA, it was then possible to examine the distribution of newly-synthesized histones. Figure 6 shows that no significant deposition of this histone occurs on new DNA during the lifetime of the closely packed nucleosomal structure. This is in agreement with the results reported for fixed chromatin (Jackson et al., 1976; Seale, 1976a; Hancock, 1977; Hancock, 1978). Our data do not rule out the possibility that new histones are deposited on the unpackaged new DNA strand at normal spacing. However, previous results with fixed chromatin would appear to make this unlikely. 46 Acknowledgments We thank Thomas D. Sargent for the gift of the 32 P-labeled PBR322 markers, and for helpful discussion of this manuscript. 47 References Birnie, G.D. (1978) Methods in Cell Biology 18, 23-29 . Birnie, G.D., Rickwood, D., & Hell, A. (1973) Biochim . Biophys. Acta 331, 283-294. Bonner, W.M., & Laskey, R.A. (1974) Eur. J. Biochem . 4 6 , 83-88. Burgoyne, L.A., Mobbs, J.D., & Marshall, A.J. (1976) Nucl. Acids Res. 3, 3293-3304 . Burke, J.F., & Pearson, C.K. (1976) Biochem. Soc . Trans. 4, 787-789. Fakan , S . , Turner, G.N., Pagano, J.S., & Hancock, R. (197 2) Proc. Natl. Acad . Sci. USA 69, 2300-2305. Hancock, R. (1974) J. Mol. Biol. 86, 649 - 663. Hancock, R. (1977) Methods in Cell Biology 16, 459-46 6. Hancock, R. (1978) Proc. Natl. Acad. Sci. USA 75, 2130- 2134. Hildebrand, C.E., & Walters, R.A. (1976) Biochem. Biophys. Res. Comm. 73, 157-163. Jackson , V. , Granner, D. K., & Chalkley , R. (1975) Proc. Natl. Acad. Sci. USA 72, 4440-4444. Jackson, V., Granner, D. , & Chalkley , R. (1976) Proc. Natl. Acad. Sci. USA 73, 2266-2269. Keppel, F . , Allet, B., & Eisen, H. (1977) Proc . Natl. Acad . Sci. USA 74, 653-656. Laskey, R.A., & Mills, A.D. (1975) Eur. J. Biochem. 56, 335-341. Leffak, I.M . , Grainger, R., & Weintraub, H. (1977) Cell 12 , 837-845 . Levy, A., & Jakob, K.M. (1978) Cell 14, 259-267. Levy, A., Jakob, K.M., & Moav, B. (1975) Nucl. Acids Res. 2, 2299-2303 . 48 Maniatis, T., Jeffrey, A., & van deSande, H. (1975) Biochemistry 14, 3787-3794. Marshall, A.J. , & Burgoyne, L.A. (1976) Nucl. Acids Res. 3, 1101-1110. Murphy, R.F., Wallace, R.B . , & Bonner, J. (1978) Proc. Natl. Acad. Sci. USA 75, 5903-5907. Murphy, R.F., Pearson, W.R., & Bonner , J. (1979) Nucl. Acids Res., in press. Noll, M. (1967) Nature 215, 360-363. Noll, M., & Kornberg, R.D . (1977) J. Mol. Biol. 109, 393-404. Noll, M., Thomas, J.O., & Kornberg , R.D . (1975) Science 187, 1203-1206. Rickwood, D., Hell, A., Malcolm, S., Birnie, G.D., MacGillivray, A.J., & Paul, J. (1974) Biochim, Biophys . Acta 353, 353-361. Riley, D., & Weintraub, H. (1979) Proc. Natl. Acad. Sci. USA 76, 328-332. Savitzky, A., & Golay, M.J.E . (1964) Anal. Chem. 36, 1627-1639. Seale, R.L. (1975) Nature 255, 247-249. Seale, R.L. (1976a) Proc. Natl. Acad. Sci. USA 73, 2270-2274. Seale , R.L . (1976b) Cell 9 , 423-429. Seale, R.L . (1978a) Cold Spring Harbor Symp. Quant. Biol. 42, 433-438 . Seale, R.L. (1978b) Proc . Natl. Acad . Sci. USA 75 , 2717-2721 . Seale, R.L . & Simpson, R.T. (1975) J. Mol. Biol. 94, 479-501. Weintraub, H. (1976) Cell 9, 419-422. PART V Chapter 6. Computer programs for analysis of nucleic acid hybridization, thermal denaturation, and gel electrophoresis data 50 COMPUTER PROGRAMS FOR ANALYSIS OF NUCLEIC ACID HYBRIDIZATION, THERMAL DENATURATION, AND GEL ELECTROPHORESIS DATA Robert F. Murphy, Wi lliam R. Pearson• and James Bonner Division of Biology, California Institute of Technology, Pasadena, CA 91125, USA •Department of Microbiology, Johns Hopkins Medical School, Baltimore, MD 21205, USA 51 ABSTRACT Computer programs for the analysis of data from techniques frequently used in nucleic acids research are described. In addition to calculating non-linear least-squares solutions to equations describing these systems, the programs allow for data editing, normalization, plotting and storage, and are flexible and simple to use. Typical applications of the programs are described. INTRODUCTION The increasing complexity and volume of data being generated in biochemical and biophysical experimentation, and the proliferation of min i-computer systems, has created a need for portable, interactive data storage and analysis programs. This paper describes non-linear least-squares fitting programs which have been used for the analysis of data from gel electrophoresis (1-4), DNA-DNA and RNA-DNA hybridization (5-7), and DNA, chromatin and protein-DNA thermal denaturation (8-10). These programs have been implemented on a PDP-11 mini-computer system, and are written in FORTRAN for ease of transfer to other computer systems. The programs require approximately 16,000 16-bit words of memory and a disk mass storage device (such as flexible disk). Mini-computers capable of running this data analysis system are currently available for less than $5,000. In this paper we will first describe three different programs in the package, and then discuss the reliability and significance of the parameter estimates calculated by the progra ms. • 52 fli.Q.QRAM DESCRIPTIONS The three programs in the least-squares fitting package are COTFIT, for analysis of nucleic acid hybridization and denaturation data, GELFIT, for determination of the positions and areas of Gaussian curves fit to gel electrophoresis data, and MELSMR, for removal of noise from closely-spaced data. These programs use a common data file format, which allows interaction of the various programs for uses other than those described above. For example, CsCl density gradient data could be entered .w ith COTFIT, smoothed with MELSMR, and then analyzed with GELFIT. COTFIT COTFIT is a non-linear least-squares fitting program developed from the NNNBAT program described by Pearson, Davidson and Britten (11), which was in turn based on the FINGER program of Britten, Graham and Neufeld (12). The program accepts English commands to control data entry and fitting. Data may be read from disk files or entered at the terminal. In addition to least-squares fitting, COTFIT provides general facilities for entering and editing data, and offers a variety of options for plotting and printing the curves calculated from the data. Initial parameter estimates are improved by a modification of the method of Marquardt (13). Parameter values which would produce undefined function values are detected without causing arithmetic errors. This enables the fitting routine to try a broad range of possible parameter values without causing program termination. As an additional option, files containing function values (with or without specified errors in the parameters) over a given interval can be generated. This feature is useful for displaying the functions under various conditions. The functions described by Pearson, Davidson and Britten (11) have been modified, and a new function has been added. The NNNBAT function names and the corresponding COTFIT names are listed below . 53 N COTFIT NNNBAT DESCRIPTION 1 2 3 FINGER DIGEST DRIVEN FINGER WHATOR NUFORM 4 5 EXCESS MELTFN ------ second order DNA renaturation. variable order renaturation. tracer/driver reaction with different nucleation rates. first order renaturation. thermal denaturation function. EXCESS The MELTFN function is H(T) = I + m T -(x-Tmi) 2 I: Fi{0.5 + --/ exp [------~--]dx} 1-ir Tmi 2Si 1=1 (1) where His the hyperchromicity at temperature T, I is the initial hyperchromicity, m is the number of components, Fi, Tmi, and Si are the hyperchromicity, transition midpoint and transition width (equivalent to the standard deviation of a Gaussian curve) of component i. The function is a normalized form of the error function, which is the integral of the Gaussian distribution . The program uses a polynomial approximation to this integral (14). The addition of this function allows direct fitting of thermal denaturation data, as opposed to the more common use of the Gaussian function to fit derivatized data. Although the two methods are theoretically identical, in practice the new approach eliminates errors introduced by the derivatization and smoothing process, and requires significantly fewer data points to determine component parameters. This is especially useful for analyzing DNA melts assayed by hydroxyapatite or filter binding (R.B. Wallace, G. Schaeffer , T. Hirose, K. Itakura, R. F. Murphy and J. Bonner, submitted to Nucleic Acids Res). GELFIT The GELFIT program uses a set of commands which is compatible with that of the COTFIT program, and fits Gaussian curves to integer data with evenly-incremented X values. The program can calculate the molecular .weight of the species in a given band using a polynomial approximation to standards data specified by the user. The simplicity of the function being fitted allows the program to make initial estimates of the number and position of the bands in a given gel, and then proceed with fitting. The restriction to Y-only integer data was added to minimize data storage spa c e and computation time in view of the 54 large numbers of data points commonly collected during gel scanning in our laboratory. The fitting method is similar to that used by COTFIT. In order to reduce program size, a band matrix (15) is used in place of the symmetric triangular matrix COTFIT uses. The number of adjacent curves whose values are allowed to affect an individual curve's parameter estimates can be adjusted at run-time by the user. MELSMR The MELSMR program smooths and/or derivatizes data having a constant X increment between points using the method of Savitzky and Golay (16-18). The degree to which the data are smoothed is controlled by the user. Digital data which ha~been collected from instruments such as spectrophotometers frequently contain fluctuations in the less significant digits. MELSMR provides a means of reducing or eliminating this noise. llS!.ll..ll A.HQ DISCUSSI O~ While these programs can significantly shorten the time required to analyze gel and melt data, and are essential for accurate measurement of nucleic acid hybridization rates and component amounts , the significance of the calculated parameter estimates must not be overestimated. This section addresses three issues enc ountered in fitting nucleic acid data with least squares programs : 1) The significance of the exponent in S1 nuclease-assayed reassociation data analysis; 2) the effect of using equation (1) to analyze melt data; and 3) the reprodu c ibility of me lt data parameter estimates. Analysis of nuclease-assayed reassociation data Morrow (19) and Sm i th , Britten and Davidson (20) have analyzed the kineti c s of DNA-DNA reassociation assayed by the single-strand-specific S1 nuclease of mtt&.i.lll.l.S. ooz.ae_. They concluded that data from S1 nuclease assayed renaturations were best fit using the equation (2) where Sis the S1 nuclease sensitive (single stranded) DNA NT concentration , Co is the total DNA NT concentration, tis time, k 55 Table 1 . Parameters for equation (3) for data of Sala-Trepat et al (6) DRIVER* liver c-DNA+ N++ u RSA 16 -0.0482 0. 1154 kidney RSA 14 hepatoma RSA liver kidney hepatoma F K 1 . 04 90 0.8782 -0.0389 0. 1240 17 AFP AFP AFP 0.001104 0.000466 n 0.440 0.965 %E+++ 3.021 2.257 1 . 0414 0.8543 0.001160 0.000472 0.440 0.976 3.760 3.033 -0.0713 0.0885 1 . 064 4 0.8980 0.001109 0.000506 0.440 0.903 2.404 1. 748 14 -0.0543 0.0628 1.0510 0.9280 0.001344 0.000736 0.440 0.745 2.204 1.718 12 -0.0643 0.0925 1.0542 0.8884 0.001317 0.000498 0.440 1. 000 2. 128 -0.0663 0.0742 1.0593 0.9110 0.001420 0.000673 0.440 0.846 2 . 932 2.422 17 3.310 • sheared to 300-400 nucleotides. The first values are with n fixed, the second with n allowed to vary. + 1000-2200 nucleotides (7). RSA=rat serum albumin. AFP=alp~a feto-protein. ++ N = number of points. +++ %E = root mean square error (RMS) divided by data mean. --- is the rate constant which would be observed if the reaction were assayed on HAP, and n was found to be 0.44 (19) or 0 . 453 (20) for The deviation from driver and tracer DNAs of the same size. second order kinetics indicated by a value of n less than 1 was attributed to the lowered reactability of the single strand regions of partial duplexes relative to free single strands. Smith, Britten and Davidson (20) observed that while no simple physical meaning can be associa t ed with the exponent n, equation (2) is useful for data reduct i on . consisting of For incomplete reactions, or reactions multiple components, COTFIT uses the DIGEST function m F(t) = U + I: Fi( 1 + KiCot)-n ( 3) i=l where U is the fraction unreacted (single-stranded) at infinite time, m is the number of components, and Fi and Ki are the fraction and rate for component i. Table 1 contains the best-fit parameters of this function for the data of Sala-Trepat et al (6 ) 56 Table 2. Comparison of melt fitting methods using data from Wallace et al (8) Difference Mean StdDev IF(B) - F(A) I ITm(B) - Tm(A) I IS(B) -S(A)I 0.0167 ,. 1 2. 1 0.0285 2.9 ,. 1 42.0 1.99 33.0 Parameter Mean StdDev Maximum S(A) S(B) 6.2 4.4 5.23 2.4 1. 6 1.32 0.41 1,. 8 7.7 7. 41 1. 92 JE(A) JE(B) 1. 22 JMean relative to A Initial absorbances were 0.9-1.3 A260 unit at 25°c. Data were collected every 0.4° while heating at 0.25 5/min. A. Parameters for Gaussian curves fit to 15-point cubic-quartic first derivative of absorbance data (using GELFIT). B. Parameters for equation (1) fit to the absorbance data normalized to fraction hyperchromicity (using C0TFIT). when the exponent n is fixed at 0.44 or allowed to vary. The unfixed exponents vary from 0.745 to 1. This shift toward second-order kinetics may be due either tothe different driver and tracer lengths used, or to the interrupted nature of the albumin and AFP genes in rat DNA (T.D. Sargent, J.R. Wu, J. Sala-Trepat, R.B. Wallace, T. Reyes, and J. Bonner, manuscript in preparation). In either case, the results demonstrate the need for careful determination of the exponent for individual S1 assayed experiments, since the calculated K values may vary by greater than 2.5 fold. Values of n significantly different from the expected value may indicate inaccurate tracer and driver length determinations or other systematic error. Analysis of DNA and chromatin thermal denaturation data Since many estimates of chromatin and DNA melting component parameters have been made by fitting Gaussian curves to derivatized data, we have compared this method to the use of the C0TFIT MELTFN function. Data from our previously published melts of various chromatin and nucleosome samples (8), which had been derivatized with MELSMR and fit with GELFIT, were re-fit using COTFIT. As Table 2 shows, the two methods yield similar results, the Tm's differing by an average of only 2J. However, COTFIT yields an average error almost five times lower than that 57 ----- Table 3. Rat liver chromatin melting transitions Sample 1 Sample 2 Sample 3 Hean StdDev JStdDev I 0.0068 0.0033 0.0047 0.0049 0.0018 36 F1 Tm1 S1 0.0989 65.5 9.5 0. 1034 60.7 7.9 0 . 1076 56.4 6.2 0.1033 60.9 7.9 0.0044 4.5 1. 6 4 7 F2 Tm2 S2 0.0789 76.9 5.8 0.0978 76.8 5.2 0.0508 69. 1 3.7 0.0758 74.3 4.9 0.0237 4.5 1. 1 F3 Tm3 S3 0.0742 77.9 6.0 0.0811 77.4 9.7 0.0652 76.4 8.4 0. 0735 77.2 8.0 0.0080 0.8 1.9 F4 Tm4 S4 0.0660 82.4 0.0431 82.6 1.9 0. 1188 79.8 4.2 0.0760 81. 6 2.7 0.0389 1. 6 1.3 51 1. 09 1. 62 1.46 0.30 22 2. 1 1. 67 21 31 6 23 11 1 24 2 46 Melts were performed in 0.25 mM ED A pH 8 . Initial absorbances were 0 0.7-1.5 A268 units at 25 C. Data were collected from 25-95 C every 0 . 4 while heating at 0 . 25°/min, and fit to equation (1) using COTFIT. The average percent standard deviations in F, Tm, and Sare 24, 4, and 29, respectively . 6 34 >u 1~ 0 a:: I u a:: w 0.. >I o.__ ___.......___~------'-------------~~ 20 50 80 TEMPERATURE °C Figure 1. Fitting of DNA melting transitions using MELTFN. From left to right are rat liver DNA and oligomers A (every 4th point shown), C (every other point shown), and B (every other point shown) (see Table 4). 58 Table 4. DNA melting transitions I I. Unsheared Rat liver DNA. A. 0.25 mM EDTA pH 8 B. 0.01 X Fl Tm1 S1 1.1-1.6 A260 units/ml, JE 0.25 0 c/min. Mean (5) StdDev %StdDev 0.4 °c/pt, 30-60 °c. -0.0011 0.3369 43.6 3.86 0.0108 0.0040 3 . 9 0.47 8.9 13. 1 3.2 378 1. 88 0.81 43 . 4 SSC pH 7 Mean (2) StdDev '.'StdDev 0.4 °c/pt, 35-75 oc. 0.0064 0.3554 58.2 0.0008 0.0069 0. 1 12.6 2.0 0.2 1.83 0.01 0.4 6.09 0.28 4.6 II. Synthetic oligonucleotides. 1 M NaCl 0.01 M PB, 0.5 0 c/min. oc. A. CCGAATTCGG GGCTTAAGCC 0 . 77-0.78 A260 units/ml, 0 . 1 °C/pt, 25-65 0.0049 0.2808 50.3 9.28 Mean (2) 3. 18 StdDev 0.0019 0.0223 0.4 0 . 49 0.39 38. 1 0.8 %StdDev 5.2 12 7.9 GGATCACCGCC B. CCTAGTGGCGG 0.72-0.77 A260 units/ml, 0.2 °C/pt, 40-70 oc. 0.0151 Mean ( 3) 0. 1817 58.6 7 . 26 2.06 0.0012 StdDev 0.0050 0.2 0.40 0.56 %StdDev 7.9 2.7 0.3 5.5 27 CATGAATTCATG 0.2 °c/pt, C. GTACTTAAGTAC 0.55-0.57 A260 units/ml, 25-70 oc. 0.0047 Mean (2) 0.2396 49.7 8.34 2.29 0.0006 0 . 0118 StdDev 0.3 0.32 0.29 12.8 3,8 '.'StdDev 4. 9 0.6 13 produced by GELFIT. Thus, although the MELSMR/GELFIT method has the advantage of ease of visual interpretation of plots, the COTFIT method produces more accurate parameter estimates. To test the reproducibility of this function, data from melts of rat liver chromatin, rat liver DNA, and synthetic oligonucleotides were fit using COTFIT (Tables 3 and 4 and Figure 1). The rat liver chromatin and DNA melts are of differen t sample preparations, and the oligonucleotide melts are of different samples from the same preparation. All melts were from separate runs. Some fluctuations in Tm resulting from differences in buffer concentration can be seen for the 0.25 mM EDTA melts, but the standard deviation in Tm is still less than 10%. The deviations for the other melts are much smaller. The closeness of the data and fit is in accordance with the shape predicted by theoretical treatments of nucleic acid melting 59 (21-24), although there is no immediate correlation between the parameters of equation (1) and the physical parameters of the system. The agreement of the rat DNA melt data with equation (1) is probably due to the variation in nucleotide composition of rat DNA, and a resulting combination of a Gaussian distribution of small transitions (25). Deviations from the fitted curve are more apparent for the oligonucleotide melts, as might be expected. In light of the ease of estimation of the parameters of equation (1), our results lend support to its use in fitting nucleic acid thermal denaturation data, especially for comparative purposes. CONCLUSlQHS We have described a set of flexible, interactive programs for the analysis and storage of biochemical data. The ability of the programs to accept English commands and prompt the operator for needed information allows even an inexperienced computer user to analyze a reassociation curve or gel profile in under an hour. In addition to 'the analysis of data from nucleic acid hybridization, thermal denaturation, and gel electrophoresis (1-10), the programs may be used for a number of other applications, such as resolution of components in velocity and equilibrium density gradients and the determination of rate constants for enzyme reactions. The programs also provide a framework for the development of other data analysis systems. The programs described in this paper are available from the authors on a variety of machine-readable media. ~OJil.E.~IS We thank T. D. Sargent, J.W. Posakony and R. B. Wallace for helpful discussion and critical review of this manuscript . This research was supported by U.S. Public Health Service Grants GM-13762 and GM-00086. 60 REFERENCES ,. Garrard, W.T., Pearson, W.R., Wake, S.K., and Bonner, J. l1974) Biochem. Biophys. Res. Comm. 58, 50-57. 2. Murphy, R.F., and Acta 405,62-66. Bonner, J. 3. Gottesfeld, J.M., Murphy, R.F., and Bonner, J. Nat. Acad. Sci. USA 72, 4404-4408. (1975) Proc. 4. Murphy, R.F., Wallace, R.B., and Bonner, J. Nat. Acad. Sci. USA 75, 5903-5907. (1978) 5. Wallace, R.B., Dube, S.K., Science 198, 1166-1168. 6. Sala-Trepat, J.M., Sargent, T.D., Sell, S., and Bonner, J. l1979) Proc. Nat. Acad. Sci. USA 76, 695-699. 7. Sala-Trepat, J.M., Dever, J., Sargent, T.D., Thomas, K., Sell, S., and Bonner, J. (1979) Biochemistry, in press. 8. Wallace, R.B., Sargent, T.D., Murphy, R.F., and Bonner, J. l1977) Proc. Nat. Acad. Sci. USA 74, 3244-3248. 9. Bonner, J., Wallace, R.B., Sargent, T.D., Murphy, R.F., and Dube, S.K. (1978) Cold Spring Harbor Symp. Quant. Biol. 42, 851-857. 10. Bakke, A.C., Wu, J.R., and Bonner, J. Acad. Sci. USA 75, 705-709. 11. Pearson, W.R., Davidson, E.H., and Nucleic Acids Res. 4, 1727-1737. 12. Britten, R.J., Graham, D.E., and Neufeld. B.R. (1974) in Methods in Enzymology (L. Grossman and K. Moldave, Eds.), Vol. 29E, pp. 363-418, Academic Press, New York . 13. Marquardt, D.W. 431-441. 14. Gautschi, W. (1972) in Handbook of Mathematical Functions (M. Abramowitz and I.A. Stegun, Eds.), pp. 295-329, Dover Publications, New York. 15. Martin, R.S., and Wilkinson, J.H. (1971) in Linear (J.H. Wilkinson and C. Reinsch, Eds.), pp. Springer-Verlag, New York. 16. Savitzky, A., and Golay, 1627-1639. (1963) J. (1975) and Soc. M.J.E. Biochim. Biophys. Bonner, J. (1978) Indust. (1964) Appl. Anal. ( 1977) Proc. Britten, R.J. Proc. Nat. ( 197 7) Math. 11, Algebra 50-56, Chem. 36, 61 17. Steinier, J., Termonia, Y., and Deltour, Chem. 44, 1906-1909. 18. Madden, H.H. 19. Morrow, J. 20. Smith, M.J., Britten, R.J., and Davidson, E.H. Nat. Acad. Sci. USA 72, 4805-4809. 21. Zimm, B.H. 22. Applequist, J., and Damle, V. 2719-2721. (1963) J. Chem. Phys . 39, 23. Applequist, J., and Damle, V. 87, 1450-1458. (1965) J. Am . Chem. Soc . 24. Abzel, M.Ya. (1973) Phys. 25. Abzel, M.Ya. 101-105. (1979) (1978) Anal. Chem. J. (1972) Anal. 50, 1383-1386. (1974) Ph.D. Thesis, Stanford University. (1960) J. Chem. Proc. Phys. Rev. Nat. (1975) Proc. 33, 1349-1356. Lett. 31, 589-593. Acad. Sci. USA 76, 62 APPENDIX A GELFIT, a non-linear, least-squares Gaussian curve fitting program 63 The following program has been used for determination of peak positions and areas for polyacrylamide gel scans (Chapters 1, 2, and 4), thermal denaturation derivative profiles (Chapter 3), and density gradient distributions (Chapter 5). In this listing, input/output related routines, some of which may be computer system specific, have been omitted. Most of these are part of a library of subroutines for handling of standardized data files. The data file format used allows access by all programs in the system to files created by any program. hybridization, thermal denaturation, and ~el C electroPhoresis data' C Submitted to Nucleic Acids Research, C C MODULES USED C GELFIT V04-10 HAIN PROGRAM C GELCMD V02-05 INTERPRETS COMMANDS C C PLTLIM V00-01 GET DEFAULT PLOT LIMITS COMNUH V00-04 GET COMMANDS USING GTLIN <SYSLIB) C COHIND V00 -03 USED BY GELCMD C IMXMIN VOl-02 DETERMINES MA X I MIN OF INTEGER ARRAY C C HELP V04 PRINTS A HELP FILE FROM DEVICE SY: SORTG voo C SORTS CURVES C FTGAUS V04-04 MAIN FITTING ROUTINE C GELFUN voo CALCULATES GAUSSIAN FIT VALUE !BANDT V02-0S FORMS DECOMPOSITION BAND MATRIX C BANSOL V02-05 GENERATES PARAMETER CHANGE VECTORS C GELLST V02 C TYPES OUT NEW FIT AND MW ' S GELPLT VOl-01 PLOT DATA AND FIT ON TERMINAL C C TTPLOT V03-01 US ED BY GELPLT GUESSG V00-04 MAKE STARTING PARAMETER GUESSES C C GENERD V01 GENERATE DATA FROM PARAMETERS C GAUS voo GENERATES GAUSSIAN RANDOM NUMBERS C C BIODAS LIBRARY SUBROUTINES C C REMARK V02-05 PRINTS MESSAGES ON UNIT LWD ERROR C VOO PRINTS FORMATTED ERROR MESSAGES OPEN C V01 CONNECTS I/0 DEVICES TO LOGICAL UNITS IRDYO C VOl-03 READS A BIODAS FORMAT DATA FILE C IFMTYP VOl-01 DETERMINES VARIABLE FORMAT DATA TYPE C LRDY1 V01 READS REAL DATA, CONVERTS TO INTEGER IRDP C VOl READS PARAMETERS ONLY (STANDARDS FILE) C NTITLE VOl-01 UPDATE FORMATTED TITLE FOR OUTPUT FILE C C CONNECTION OF FILES TO LOGICAL UNITS IS HANDLED BY THE C SUBROUTINE 'OPEN', AND IS DONE ONLY IN THIS MODULE C AND IN 'HELP,' C C THIS PROGRAM WILL RUN 'AS IS' <IF OVERLAID> ON ANY PDP-11 C WITH AT LEAST 28K WORDS MEMORY RUNNING UNDER RT-11 VOJB, (: PROGRAM GELFIT C V04-10 28-JUN - 79 C C R,F,HURPHY CREATED 23 - AUG - 76 C DIVISION OF BIOLOGY C CALIFORNIA INSTITUTE OF TECHN OLOGY C PASADENA, CA 91125 C (2 13) 795 - 6811 EXT,1915 C C FITS GAUSSIAN COMPONENTS TO VARIABLE FORMAT INTEGER DATA FILES C USING BAND MATRICES TO MINIMIZE STORAGE REOUIREMENTS, C C USERS OF THIS PROGRAM SHO ULD CITE: C MurPh~, R,F ,, Pearson, W,R ,, _ & Bonn er, J, 11979) C •compu t er Programs for analysis of nucleic acid C GELFIT.FOR C COMMON/V0410/V0410 INTEGER V0410 COMMON/PRGNAM/PRGNAMC16) BYTE PRGNAM COMMON/FILCOM/OUTFIL<16),INPFIL<16),LSTFILC16>, *PLTFILl16>,STDFIL116) BYTE OUTFIL,INPFIL,LSTFIL,PLTFIL,STDFIL COMMON/LOGDEV/LRD,LWD,LFD INTEGER LRD,LWD,LFD COMMON/DATA/TITLEIBO),MINMAX(4),NOPTS,HAXPTS,IY1600) BYTE TITLE INTEGER NOPTS,MAXPTS,IY REAL MINMAX COMMON/CURVES/NOCURV,NOPAR,MAXPAR,PC9O) INTEGER NOCURV,NOPAR,MAXPAR REAL P COMMON/FITPAR/NIT,NCIT,RHSOU,DELOU,RATMAX,NE!GHB,NFPAR, * MAXBND,XDEL,NFREE,OCOND,RMS,CHISO,DELMX INTEGER NIT,NCIT,NEIGHB,NFPAR,MAXBND,NFREE,OCOND REAL RMSOU,DELOU,RATMAX,XDEL,RMS,CHISO,DELMX COMMON/PARCAL/PARCAL LOGICAL PARCAL COMMON/FIXCOM/FIXED,UNFIXD,FIXARRC3O) BYTE FIXED,UNFIXD,FIXARR COMMON/RANCOM/IRANV1,IRANV2 INTEGER IRANV1,IRANV2 COMMON/PLOTCM/XPMIN,XPMAX,YPMIN,YPMAX,NXPLOT, * NYPLOT,NYPERX,LOGFLG,CRTPLT LOGICAL*! LOGFLG,CRTPLT INTEGER NXPLOT,NYPLOT,NYPERX REAL XPMIN,XPMAX,YPMIN,YPMAX COMMON/STDCOM/NOSTD,MAXSTD,LOGSTD,STDCB) INTEGER NOSTD,MAXSTD r ( TH E RT - 11 VERSION OCCUPI ES 15,5K WORDS , EXCLUDING BUFFER S > C ro INCREASE THE Nl~BER OF CURVES , CHANG E lHr DIM ENSIONS CON P AND rIXARR, 1:HANGE lHE DATA STATEMFNI fOR HAXPAR, C AND rHANGE. THE DIMENSIONS ON PARlDX,PARIDY,DELTA,OLDP AND IMAP C IN MODULE 'FT GAUS ' <OR REMOVE THE COMME NT INDICATOR C FROM THE COM MONS BELOW AN!t CHANGE THE DIMEN SIONS; SEE C 'FTGAUS ' FOR EXPLANATION OF EFFECT,) C TO INCREASE THE NUMBER OF NEIGHBORS, CHANGE THE DIMENSION C FUR WRKSPC, AND CHANGE MAXBND, C TO INCREASE THE NUMBER OF POINTS, CHANGE THE DIMENSION CON IY, AND CHANGE MAXPTS, C C TO SAVE CORE, AN OVERLAID PROGRAM CAN BE GENERATED. C OVERLAY REGION MODULES C ROOT GELFIT,REMARK,ERROR C tl GELCMD C tl IRDYO,IFMTYP,IRDP,NTITLE,IWDYO C t1 FTGAUS,IBANDT,BANSOL C t1 OPEN,HELP,SORTG,GELLST C t1 GELPLT,TTPLOT C t1 GUESSG,GENERD,GAUS C t2 COMNUM C t2 PLTLIM C t2 IMXMIN C t3 COMIND C t3 GELFUN +:' "' INTEGER COHAND DATA LRD,LWD,LFD DATA MAXPTS 16001 DATA MAXF·AR /90/ DATA MAXBNI• /450/ / 8/ DATA MAXSTD DATA FIXED,UNFIXD I,, I, I , ', , , , I I * ' L ' ,'S ' , ' T',' ' , ' , , , ', ' ' , ' - ' / DATA PLTFIL /'L','P', ' :','D ' ,'A', ' T','A','• ' • * 'f'', 'L', 'T',, , , , , , I I, I I , I I I / 'L','P',': ', ' D' ,'A ',' T' ,'A','• ' • '[I', 'A ' ' ' T', , DATA LSTFIL * DATA NOPTS, NOPAR, DEBUG /0,0,0/ DATA RHSOU,DELOU,RATMAX / 0,,0,,0, / DATA OUTFIL /'D','K',':', ' D' ,'A', 'T', 'A ' ,'• ' • 'I '/ C PRINT PROGRAM NAME AND VERSION NUMBER, CALL REMARKIPRGNAM> COMAND = 0 CHAIN GELFIT LOOP START, GET COMHANn, 5 CALL GELCMDICOHAND> IF I COMAND,E0,26> GOTO 260 C C C SETTING THESE VARIABLES TO ZERO INITIALZES THE RANDOM C NUMBER GENERATOR, DATA IRANV1,IRANV2 /0,0/ C /'f.',, /5,7,2/ ~'•'• /'G','E','L','F','I','T', *' ' , ' V','0 ' , ' 4', '-' ,'1', ' 0',' DATA PRGNAM C INITIAL VALUE DATA STATEMENTS C C COPEN RETURNS -1 IF FILE OPENED, OTS ERROR t OTHERWISE, INTEGER OPEN EXTERNAL GELFUN C C LOGICAL LOGST[r REAL STD er COHHON / PARTDX/PARTDXl901 er REAL PART[rX er COHHON/PARTDY/PARTDYl901 C! REAL PARTDY COHHON/DELTA /DEL TAl901 C! REAL [IELTA C! COHHON/OLDP/OLDPl901 C! C! REAL OLIW er COHHON /I HAP/IHAP1901 er INTEGER IMAP COHHON/G UESCH /YF ILT,ZFILT REAL YFILT,ZFILT COHHON/VFRHAT/VFRHATC10) INTEGER VFRHAT COMMON / DEBUG / DEBUG INTEGER DEBUG C CHARACTER ARRAYS BYTE HLPTHP180) EQUIVALENCE I HLPTHPl11,WRKSPCl1)) C NUMERICAL ARRAYS REAL WRKSPCl900) FIT CALL SORTGIP,NOPAR,1) ! 1 FOR ORDERING FOR OUTPUT CALL IHXHIN(IY,NOPTS,IYMAX,IYMIN> HINMAXl3) = FLOATIIYHINJ HJNHAXl4) = FLOATIIYHAX> IFIOPENILFD,OUTFIL, "DAT ' ,'NEW ' ,'NC'l,GE,O> GOTO 5 CALL NTITLEITITLE,OUTFIL,PRGNAH,OJ CALL IWDYOILFD,TITLE,NOPTS,NOPAR,HINHAX,VFRHAT ,IY ,P> CLOSE IUNIT=LFD I GOTO 5 CLOSE IFIOPEN<LFD,STDFIL,'DAT','RDO ' , ' NC'),GE,OJ GOTO 5 CALL IRDPILFD,WRKSPC,NOSTD,STD,MAXSTD) CLOSE IUNIT=LFD > GOTO 5 STANDARDS IFCOPENILFD,INPFIL, ' DAT ', 'RDO','NC ' l,GE,O> GOTO 5 CALL IRDYOILFD,TITLE,NOPTS,NOPAR,HINMAX, f.VFRMAT,IY,P,HAXPTS,MAXPAR> CLOSEIUNIT=LFDI NOCURV = NOPAR/3 IFINOPTS,GT,1) XDEL = IHINMAX121-MINHAXl1))/INOPTS-1> NEIGHB = -1 RATMAX = 0, RHSOU = 0, [IELOU = 0, 0001 IF I NOPAR,LE,O) GOTOS DO 12 I=l,NOPAR FIXARRII) = UNFIXD CONTINUE CALL SORTGIP,NOPAR,O> ! 0 FOR ORDERING BY XMAX GOTO 5 C 106) C PLOT CALL FTGAUS<WRKSPC> CALL SORTGIP,NOPAR,0) C DON ' T UPDATE REVISION NUMBER UNLESS SOME ITERATION~ WERE DONE I =4 IF<NIT,LE,Ol I= 0 CALL NTITLEITITLE,OUTFIL,PRGNAM,l) IFIOPEN<LFD,LSTFIL, ' LST','NEW ' , ' CC'J,GE,O> GOTO 5 CALL GELLST CLOSECUNIT=LFt•> GOTO 5 50 C C 105) C 40 C C 104) C C 1031 C 30 C 12 C (02) C 20 INPUT CALL HELPIHLPTHP) GOTO 5 10 C HELP IF ICO HAN[l - 23) 6,230 , 240 Gora ( 10 , ~0 ,3 0 ,4 0 , 50 ,60) COHAN[I C C ( () 1 ) C 6 \Jl 0\ 1231 GENERATE CALL GUESSGIIY,NOPTS,HINHAXl11,XDEL,INTIYFILTl,ZFILT, *P,NOPAR,HAXPAR) DO 232 I=l,NOPAR FIXARRIII = UNFIXD CONTINUE NOCURV = NOPAR/3 GOTO 5 GUESS CALL EXIT END EXIT CREATED 15-HAY-78 C C C C C C C C C C C C C C C C C C C C C C C C C C COTF IT LWD OUTFIL<16) INPFIL< 161 LSTFIL<16> F'LTFIL<161 STDFIL<161 TITLE180) XHIN XHAX YHIN YHAX NOF'TS HAXPTS IY I HAXPTS I VFRHAT NOCURV NOf'AR HAXPAR P<HAXPARI NIT NCIT RHSOU COHAND RATHAX NEIGHB DELCW c urn C C C C C GELF IT Crt Lon!! Add NOLOG I 2R 2R 2R I,3R R I 21 21 2R 3R 21 21 21 21 21 21 4R,3I FI LE FILE FILE Of'T IONS ARGS /DATA/ /VFRHAT/ /CURVES/ /CURVES/ /CURVES/ /CURVES/ /FITPAR/ /FITPA R/ /FITf'AR/ /FITPAR/ /FITf'AR/ /FITPAR/ /DATA/ /DATA/ /DATA/ /DATA/ /DATA/ /DATA/ /[IATA/ /LOGDEV/ /LOGDEV/ /FILCOH/ /FILCOH/ /FILCOH/ /FILCOH/ /FILCOH/ <ARG > > >< t OF DATA POINTS >MAX VAL FOR NOF'TS >< Y DATA VALUES <OUTPUT DATA FILE FORMAT >< t OF FIT CURVES INOPAR/3) >< t OF FIT PARAMETERS >MAX VAL FOR NOPAR >< PARAMETERS >< t ITS FOR REFINE >< t CITS FOR REFINE >< RMS OUIT VALUE >< DELHX OUIT VALUE >< MAX RATIO OF HEIGHT TO HWIDTH >< t OF NEIGHBORING BANDS IN FIT > > >LAST COHHAND <NEW COHHAND t >o:oK; =O:BAD COHI =-1:AHB COH =-2:BAD SWITCHI =-3lAHB SWITCH >LOGICAL READ DEVICE >L OGICAL WRITE DEVICE ><OUTPUT FILE NAME >< INPUT FILE NAME ><LISTING FILE NAME ><PLOT FILE NAME ><STANDARDS FILE NAME <NEW TITLE AFTER POINTS I LOG,NOLOG I,3R 2R 2R 21 21 2R 13R 21 21 21 21 21 LI st, F'Lot 21 Crt,LP,LOG,NOLOG 4R,3I List.PLot FILE R Lons Add,Old FILE OPTIONS ARGS C INPUTS <HARKED > > AND OUTPUTS <HARKED < > c 26 E!{ i t C C C COMMAND OUTLINE C C GETS AND INTERPRETS COTFIT COMMAND STRINGS, C RETURNING AN INTEGER COMMAND NUMBER FOR C THOSE COMMANDS WHICH NEED PROCESSING BY THE CHAIN PROGRAM, C C R,F,HURPHY C SUBROUTINE GELCHD<COHANDI C V02-05 28-JUN-79 c------------------------------------------------------------ c-----------------------------------------------------------c GELCHD,FOR 260 C C (26) C XDEL = IHINHAXl21-HINHAX<111/INOPTS-1> GOTO 5 *WRKSPCINOPAR ♦ ll,GELFUN,NOPTS,,,,IY,,TRUE,,,TRUE,I C PUT DESIRED STANDARD DEVIATIONS INTO WORKING ARRAY, 24 0 DO 242 I=l,NOPAR WRKSPCIII = 0, IFIFIXARRIIl,NE,FIXEDI WRKSPCIII s HINHAX<31*P<II 242 CONTINUE C THE ,FALSE, INDICATES NO 'LOGX' GENERATION, C THE TWO ,TRUE,'S INDICATE INTEGER Y ONLY DATA GENERATION , C WRKSPC<NOPAR+l> IS USED FOR STORAGE OF OLD PARAMETERS C WHEN NEW PARAMETERS ARE GENERATED FOR EACH DATA POINT, CALL GENERDCHINHAX(1),HINHAX(2l,,FALSE,,NOPAR,P,WRKSPC, C C GELCHD RETURNS THE NUMBER OF POINTS TO GENERATE IN 'NOPTS', C THE START AND STOP X VALUES IN 'HINHAXlll' AND 'HINHAXC21', C AND THE RELATIVE STANDARD DEVIATION FOR ALL UNFIXED PARAMETERS C IN 'HINHAX l3 1' IIF ZERO, THE EXACT FIT IS GENERATED>, C C 124) C 232 230 C C C C 01 Hell" C 02 lnl"ut C 03 Sta ndards C 04 Close C 05 FIT C 06 Plot C 07 Outl"ut C 08 New C 09 Add C 10 Delete C 11 Modify C 12 TYl"e C 13 FIX C 14 Unfix C 15 POints C 16 Kill C 17 List C 18 Quit C 19 Bands C 20 Width C 21 Xscale C 22 Yscale C 23 GUess C 24 GEnerate C 25 DEBu!I C t COH HANT• C 60 CALL NTITLEITITLE,OUTFIL,PRGNAH,01 IFIOPENILFD,PLTFIL, ' PLT ' , ' NEW ' , ' CC'J,GE,01 GOT O 5 CALL GELPLTIWRKSPC,HAXBND* 21 CLOSE I UN IT=LF[o I GOTO 5 C 0-0-- / FITPAR/ / FITPArU /FilT'AR/ XDEL /FIXCOH/ FIXED /FIXCOH/ UNFIXD FIXARRCHAXPARl/FIXCOH/ XPHIN, XPHAX /F"LOTCH/ YPHIN,YPMAX /F"LOTCH/ NXPLOT /PLOTCM/ /PLOTCH/ NYPLOT /PLOTCM/ NYF·ERX / F"LOTCH/ LOGFLG /f"LOTCH / CRTF"LT /S TDCOH / NOSTD / Sr[ICOH / LOGSH• l f•EBUG / DEFUG NFPAR HAXBND < ,TRUE, MEANS PLOT Y VS, LOGCXl < ,TRUE, MEANS CONDENSED PLOT CORDER OF STANDARDS FIT <FIT IS TO LOG MW >< DEBUGGING FLAG <t CHARS PER LINE IN PLOT <t LINES PER PLOT <t Y DOTS PER X <FOR FIT PLOT> >< NUMBER OF FREE PARAMETERS ~<HAX PROD OF NFPAR AND NOPAR >< X CHANGE BETWEEN POINTS >ASCII VALUE TO SHOW PARAH IS FIXED > " • • • • • NOT FIXED >< FLAGS FOR FIXING PARAMS C DATA HAXCOH /26/ DATA MAXSWI /9/ DATA MAXFLG /11/ DATA COHSTR/'H','E','L','P','l ' , ' N' , ' P ' , ' U','T ' ,'S ' , INTEGER I,J,K,IARG1,IARG2 REAL RARGl, RARG2 REAL OMN,OHX,XT C PROCEDURE TYPE DECLARATION, INTEGER COMNUH C INTE GER HAXCOH,HAXSWI,HAXFLG,L STRNG,I PTR,LARGS C 'ARGTYP' f O IF COMMAND TAKES INTEGER ARGS, · o IF REAL ARGS, C 'REODAT' ,TRUE, IF COMMAND REQUIRE S PREVIOUS DATA POINTS, C 'RECTPAR' ,TRUE, IF COMMAND REQUIRES PREVIOUS PARAMS, BYTE ARGTYP(26l LOGICAL*l REODAT(26l,REOPAR<26l * * * * * * * * * * INTEGER COHAND *'T ' , ' A', ' N','D','A','R','D','S','C','L ' , COHMON/LOG nEV/LRD,LWD,LFD * ' O' ,'S','E','F','1','T','P','L ' ,'O','T', INTEGER LRD,LWD,LFD * ' 0','U','T','P','U','T','N','E','W','A', COMHON/FIL COH/OUTFIL<l6l,lNFILE<16>,LSTFIL<l6l, *'D ' ,'D','D','E','L','E','T','E','H ' ,'0', * PLTFIL<16l,STDFILC16l *'D','I','F','Y','T','Y','P','E','F','I', BYTE OUTFIL,INFILE,LSTFIL,PLTFIL,STDFIL * ' X','U','N','F','I','X','P','0','I','N', COMHON/DATA/TITLE<80l,XMIN,XHAX,YMIN,YHAX, *'l ' ,'S','K','I','L','L', ' L','I','S','T', * NOPTS,HAXPTS,IY(lOOl *'0','U','I','T','B','A','N','D','S','W', BYTE TITLE *'I','D','T ' ,'H','X ' ,'S','C','A','L','E', INTEGER NOPTS,HAXPTS,IY * ' Y', ' S','C','A','L','E','G','U','E','S', REAL XHIN,XHAX,YHIN,YHAX *'S','G','E','N','E','R','A','T','E','D ' , COHMON/VFRHAT/VFRMAT<lOl * ' E' ,'B','U','G','E','X','I','T'/ INTEGER VFRHAT DATA COHPTR/4,l,9,5,18,10,23,l9,26,26, COHHON/ CURVES/NOCURV,NOPAR,MAXPAR,P<60l * 30,28,36,31,39,37,42,40,48,43, INTEGER NOCURV,NOPAR,HAXPAR * 54,49,58,55,61,61,66,62,72,68, REAL P * 76,73,80,77,84,81,89,85,94,90, COMMON/FITPAR/NIT,NCIT,RHSOU,DELOU,RATMAX,NEIGHB, 100,95,106,101,lll,108,119,ll3,l24,l22, * NFPAR,MAXBND,XDEL 128,125/ INTEGER NIT,NCIT,NEIGHB,NFPAR,HAXBND DATA SWISTR/'O','L','D','A','D','D', ' L','0', ' N','G', REAL RHSOU,DELOU,RATHAX,XDEL *'C','R','T','L','0','G','N','0', ' L' ,'0', COHHON/FIXCOH/FIXED,UNFIXD,FIXARR(1l *'G','L','P','L','I ' ,'S','T','P','L ' ,'0','T'/ BYTE FIXED,UNFIXD,FIXARR DATA SW1PTR/3,1,6,4,10,10,13,11,l6,l6, COHHON/PLOTCH/XPHIN,XPHAX,YPHIN,YPHAX,NXPLOT, * 21,21,23,23,27,25,31,29/ NYPLOT,NYPERX,LOGFLG,CRTPLT C SWITCHES WHICH ARE ONLY USED BY COTFIT LOGICAL*1 LOGFLG,CRTPLT C HAVE THE COMMAND t SET TO 0, INTEGER NXPLOT,NYPLOT,NYPERX DATA SWIFLG /1,0,0, REAL XPHIN,XPHAX,YPMIN,YPHAX * 2,2,0, COHHON/STDCOM/NOSTD,HAXSTD,LOGSTD * 3,1,0, INTEGER NOSTD,HAXSTD 4 , 6,0, LOGICAL LOGSTD 5,0,0, COHHON/GUESCM/YFILT,ZFILT 6,3,0, REAL YFILT,ZFILT s,o,o, COMMON/DEBUG/DEBUG * 6,0,0, INTEGER DEBUG 7,6,0, C WHEN 'ADDFLG' IS ,TRUE,, DATA IS BEING ADDED TO PREVIOUS VALUES * 8,7,0, LOGICAL*! ADDFLG 9,7,01 C WHEN 'LPTPLT' IS ,TRUE,, LINE-PRINTER DEFAULTS ARE SELECTED, DATA ARGTYP /0,0,0,0,-l,O,O,O,O,-l, LOGICAL*1 LPTPLT -1, - 1, - 1,-1,o,-1,-1,1,-1,o, C THIS ARRAY CONTAINS THE DEFAULT FILE FORMAT, * 1,1,1,0,-1,01 DATA REQDAT /0,0,0,1,1,1,o,o,o,o, I~TEGER IFRMAT<lOl,MAXIFR * o,o,o,o,o,1,1,o,o,o, C THESE ARRAYS AND VARIABLES CONTAIN DATA FOR THE COMNUH SUBROUTI BYTE COMSTR(l28/,SWISTR(3ll,CSTRNG(81) * 1,1,1,0,0,0/ DATA REOPAR /O,O,O,l,l,0,0,0,1,l, INTEGER COHPTR<2,26J,SWIPTRC2,9l,SWIFLGC3,11) C C C C C C C C C C C C C C C C C -..J °' C C IF LAST COMAND WAS lnPul DO AN IMPLICIT NeY IFCCOMAND, E0,2,AND,NOCURV,LE,0,AND,NOPT S ,GT,01 GOTO 80 C RETURN HERE TO GET NEW COMMAND , C CONVERT CO MMAND AND ASSOCIATED SWITCHES INTO COMMAND NUMBER C AND FLAGS IN SWI FL G, 1 CALL RCTRLO LSTRNG = 0 COMAND = COMNUMCCSTRNG,LSTRNG,COMSTR,COHPTR,MAXCOM, *SWISTR,SWIPTR,MAXS WI,SWIFLG,MAXFLG,IPTRI C CHECK FOR BAD COMMAND OR DAD OPTION, IFCCOMAND,GT,01 GOTO 4 GOTO <530,520,510,5001 COHAND+4 C IPTR - -> START OF ARGUMENTS, C MAKE LARGS = LENGTH OF ARGUMENTS STRING, 4 LARGS = LSTRNG - IPTR + 1 C GET INTEGER ARGS IF NEEDED, IF<ARGTYP<COMAND>> 5,9,7 5 IFCLARGS,LE,01 GOTO 6 DECODECLARGS,2000,CSTRNG<IPTRl,ERR=SOOI IARG1,IARG2 GOTO 9 6 IARGl = 0 IARG2 = 0 GOTO 9 C GET REAL ARGS IF NEEDED, IFCLARGS,LE,01 GOTO 8 7 DECODECLARGS,2010,CSTRNGCIPTRl,ERR=SOOI RARG1,RARG2 GOTO 9 RARGl = 0, 8 RARG2 = 0, C HAKE SURE REOUIRED DATA AND CURVES EXIST, IFCREODATCCOHANDl,AND,NOPTS,LE,01 GOTO 540 9 IFCREOPARCCOHANDl,AND,NOPAR,LE,01 GOTO 550 GOTO (10,20,30,40,50,60,70,80,90,100,100,100,130,130, *150,160,160,180,190 ,200, 2 10,210,230,240,250,2601 COHAND C C <01 I H(ELPI (/LONG) - PRINT COHHAND SUHHARY C IFCSWIFLG<3,31,NE,OI RETURN 10 WRITE<LWD,10401 GOTO 1 C C (021 I<NPUT IC /ADDI CINFILE,EXT> - GET INPUT FI LE C C CHECK FOR /ADD, 20 ADDFLG = <SWIFLG(3,2>,NE,OI C SEE IF A FILENAME WAS SPECIFIED, IF<LARGS,GT,01 GOTO 21 C NO ARG SPECIFIED, PROMPT, WRITE <LWD,1010 1 READ<LRD,20201 INFILE GOTO 22 C USE ARGUMENT SPECIFIED, 21 DECODECLARGS,2020,CSTRNGCIPTRII INFILE C DEFAULT OUT PUT FIL E IS THE INPUT F ILE , * 1,1,1,1, 0,0 , 0,0 ,0, 0 , * 0,0,0,!,0,0/ DATA IFRMAT /'( 1 ' , '0 l ' , ' 81 ' , ' *' , '/ DATA HAXIFR /10/ 62 60 C CALL IMXMIN(IY,NOPTS,IYMAX,IYHINI CRTPLT = <SWIFLG<3,41,NE,0 > LPTPLT = <SWIFLGC3,91,NE,0I IFCLARGS,LE,01 GOTO 62 DECODE<LARGS,2040,CSTRNG<IPTRl,ERR=500) XPHIN,XPMAX, *YPMIN,YPMAX,NXPLOT,NYPLOT,NYPERX CRTPLT = (,NOT,LPTPLT> GOTO 66 CALL PLTLIMCXMIN,XHAX,XPHIN,XPHAXI YMIN = FLOATCIYMINJ YMAX = FLOAT<IYMAX> CALL PLTLIM(YMIN,YMAX,YPMIN,YPMAXI NYPERX = 5 C DON ' T CHA NGE OUTPUl FILE JF ADDING, 22 IFCAD DFLG I GOTO 2 4 DO 23 1=1,1 6 OU TFIL<I > = INFILE<I> 23 CONTINUE C GET DEFAUL TS FOR LIST AND PLOT FILES CA LL SET FJL C NOT ADD I NG, DELETE OLD PO INTS, NO PTS = 0 C RETURN TO MAIN PROGRAM TO READ INPUT FILE, 24 RET URN C C <031 S<TANDARDSl( / NOLOGI CSTDFIL,EXTI - GET STANDARDS FILE C 30 IF<LARGS,GT,01 GOTO 31 WRITE<LWD, 10501 READCLRD,20201 STDFIL IF(STDFIL<ll,NE, ' ' I GOTO 32 C NO FILE SPEC, CANCEL PREVIOUS STANDARDS, NOSTD = 0 GOTO 1 31 DECODE<LARGS,2020,CSTRNG<IPTRII STDFIL C FITS ARE TO LOG MW UNLESS /NOLOG IS SPECIFIED, LOGSTD = (SWIFLG<3,61,EO,OI 32 RETURN C C (041 C<LOSEI - RETURN TO WRITE OUTPUT FILE C 40 RETURN C C <051 FIT CNIT>C,NCIT> - RETURN FOR FITTING C 50 NIT= IARG1 C LET FIT IHPLY FIT 10, IFCLARGS,LE,01 NIT= 10 NCIT = IARG2 C LET FIT? IHPLY FIT?,30 IF<NCIT,LE,01 NCIT = 30 IF<NEIGHB,LT,01 NEIGHB=HINO<NOCURV-1,CHAXBND/NOPARl/3-1 1 NFPAR = <NEIGHB+1>*3 IF<NFPAR,GT,NOPARI NFPAR = NOPAR IF(CNFPAR*NOPARl,LE,HAXBND> RETURN WRITECLWD,10601 NOPAR,NFPAR,MAXBND GOTO 1 C C <061 PL<OTI (/CRT) - PLOT POINTS I FIT ON TERMINAL °' 00 OCUTPUT)(/LISTJ</PLOTl<OUTFIL ,EX TI - SLT OUTPUT r ILENAME S N<EWJ - GET NEW CURVES 80 NOPAR s 0 C MAkE SURE ADD COMMAND READS FROM TELETYPE LARGS = 0 C C (09) ACDD> - ADD TO PREVIOUS PARAMETERS C 90 IF(NOPAR,GE , MAXPAR> GOTO 560 IF<LARGS,GT,O> GOTO 93 J = NOPAR 91 WRITE(LWD, 1070> 92 READ<LRD,2010,ERR=91,END•94) (P<IJ,I=J+1,J+3J J = J + 3 IF<J,LT,HAXPAR> GOTO 92 GOTO 94 93 DECODE<LARGS,2010,CSTRNG<IPTRl,ERR=500l * <P<I>,I=NOPAR+1,NOPAR+3) J = NOPAR + 3 C C (08) C Ir<SWIFLG<3,10J,NE,OI GOTO 72 lf( S WIFLG<3,11J,NI , UI GIJlll 7 4 ff<LAF:G:J ,LE ,01 GOTO 71 flFCODE<LAF:GS ,20 20 , CS Tr.· N(3( [ ~· TRI J OUTF IL GOTO I 71 WRllE <L WD,103 01 READ(LRD,2020) OUTFIL C SETTING OUTPUT FILENAME AL S O SETS PLOT AND LIST FILENAMES, CALL SETFIL GOTO 1 72 IF<LARGS,LE,O> GOTO 73 DECODECLARGS,2020,CSTRNG<IPTR)J LSTFIL GOTO 1 73 WRITECLWD,12401 READ(LRD,2020) LSTFIL GOTO 1 74 IF<LARGS,LE,01 GOTO 75 DECODECLARGS,2020,CSTRNG<IPTRI) PLTFIL GOTO 1 75 WRITECLWD,1250> READ<LRD,2020) PLTFIL GOTO 1 70 C C (07) C IF<,NOT,LPTPLT> GOTO 64 C USE LINE - PRINTER DEFAULT S NXPLOT = 101 NYPLOT = 51 GOTO 66 IF(,NOT,CRTPLTI GOTO 65 64 C USE CRT DEFAULTS AND USE CONDENSE D PLOTTING NXPLOT = 68 NYPLO T = 16 GOTO 66 C USE TELETYPE DEFAULTS AND SET CONDENSED PLOTTING FLAG 65 NXPLOT = 61 NYPLOT = 41 CRTPLT = ,TRUE, 66 RETURN MAkE SURE DELETE, MODIFY AND TYPE ARGUMENTS ARE WITHIN LIMITS DO 95 I =NOPAR+l,J FIXARR<IJ = UNFIXD CONTINUE NOF'AR = J NOCURV = J/3 NEIGH£< = - 1 GOTO 1 <II ( ,JI - DELETE CURVES MCODIFYI <l><,JJ - CHANGE CURVES TCYPE> (11(,J) - TYPE CURVES I= IARG2 - IARG1 + 1 NOPAR = NOPAR - I NEIGHB = -1 IF<IARGl,GT,NOPARJ GOTO 104 P<IARGl) = P<IARG1 ♦ 1) FIXARRCIARGl> = FIXARR<IARGl ♦ IJ IARGl = IARG1 + 1 GOTO 103 NOCURV = NOPAR/3 GOTO 1 D<ELETEI C C <13> C (14> C FIX CIJC,JI - HOLD CURVE I PARAMETER J CONSTANT U<NFIXJ <I> <,J > - ALLOW CURVE I PARAMETER J TO VARY WRITE<LWD,1080) WRITE<LWD,1090) IARGl/3 ♦ 1, * CPCI>,FIXARRCI>,I=IARG1,IARG1 ♦ 2) C IF TYPING, DON'T GET NEW VALUES, IF(COMAND,E0,12) GOTO 113 112 WRITE<LWD,1100) READ(LRD,2010,ERR=114) <P<I>,l=IARG1,IARG1+ 2 > 113 IARG1 = IARG1 + 3 IF<IARG1,GT,IARG2J GOTO 1 GOTO 111 114 CALL ERRORC ' W','inPut error - rePeat last curve,') GOTO 112 110 111 C C (111 t (12> C 104 103 C C ( 10> C C LET 'COMAND N, CR > ' IMPLY 'COMAND N,NO CURV :CR ) ' 100 IF<C STf.:N \l .'. , rnNGJ ,NE, ,· , ' I GOTO 101 IAR1;2 = OCURV liOTO 102 C LET ' CU MAHD N<CR; ' IMPLY 'COMAND N,N .CR 101 IFC1A l<U2 , ELl,O J IAR G2 = IAR Gl C Lll ' COMAND \ CR • ' IMPLY ' CO MAND 1 , NOCUR V CR , lF<LARG S , GT, O> GOTO 102 I Arm l = 1 IARG2 = NOCURV 102 IF<IARG1,L[,0,0R,IARG1,GT,NOCURV,OR,IARG2,LT,IARG1 *, OR , IA RG 2,GT, NOCURV> GOTO 600 C CONVERT CURVE NUMBERS TO PARAMETER NUMBERS IARGl = IARG1*3 - 2 IARG2 = IARG2*3 IF<COMAND,NE,10> GOTO 110 C C C C _, 9 .- 94 (j\ '° LIISTJ IHll,NJ - LIST POINTS 163 C C < 17> C 170 C = WRITE<LWD,1110) HAKE SURE KILL I LIST ARGUMENTS ARE WITHIN LIMITS C LET •tOHAND N, <CR > ' IMPLY 'COHAND N,NOPTS <CR > ' 160 I F <CSTRNG(LSTRNGJ,NE, ' ,' ) GO TO 16 1 IARG 2 = NOPTS GOTO 162 C LET 'COHAND N<CR > ' IMPLY 'COHAND N,N <CR > ' 161 IFCIARG2 , EO,O> IARG2 = IARG1 C LET 'COHAND <CR > " IMPLY 'COHAND 1,NOPTS <CR > " I F ILARGS,GT,01 GOTO 162 IARG1 = 1 C C C READILRD,20101 XHIN,XDEL IFIXDEL,E0,0,1 XDEL = 1 , 154 WRITEILWD,1120) 155 READILRD,2000,ERR=156,END•1581 IYINOPTS+l> NOPTS = NOPTS + 1 IFINOPtS,GE,HAXPTSJ GOTO 580 GOTO 155 156 CALL ERRORl'W', ' inPut error - rePeat la~t Point, ' ) GOTO 155 158 IFINOPTS,LE,OJ GOTO 540 XHAX = XHIN f INOPTS-ll*XDEL C DO AN IMPLICIT NEW IFINOPAR,LE,01 GOTO 80 GOTO 1 153 READCLRD,20 2 0) ITITLEIIl,1=44,801 DO 153 I=l,HAXIFR VFRHATIIJ = IFRHAT<I> CONTINUE WRITE<LWD,10201 C IF NO DATA YET, DO A TT: INPUT COHHAND C HAKE SURE THERE'S ROOM FOR HORE POINT S , 150 IFINOPTS,GE,HAXPTS> GOTO 580 IFINOPTS , GT,01 GOTO 154 C IF NO POINTS ALREADY, GET TITLE, ETC, NOPTS 0 NOPAR = 0 DO 152 1=1,43 TITLEII> = ' ' 152 CONTINUE WIIDTH> IA) - SET HAX RELATIVE WIDTH C 1201 C XDEL = XDE L*RARG 2 XHIN = IXHIN+RARG11*RARG2 XHA X = IXMAX+RARG1 1*RARG2 C C GET FACTORS FOR XSCALE AND YSCALE , C 210 IFI LARGS,GT,01 GOTO 211 WRITECLWD,1230) READI LRD, 2 0101 RARG1,RARG2 2 11 IF I COHAND, E0,22> GOTO 220 C C I 21 I XI SCALEJ - SCALE X DATA 201 2 00 C IFCLARGS,GT,01 GOTO 201 WRITE I LWD, 1140) READILRD,2010) RATHAX GOTO 1 RATHAX = RARG1 GOTO 1 IFCIARG1,LT,O,OR,IARG1,GE,NOCURV> GOTO 570 NEIGH.£< = IARG1 GOTO 1 190 C BCANDSJ III - SET t OF NEIGHBORING BANDS IN FIT READILRD,20101 RHSOU WRITE<Lwr,, 1200) READILRD,2010> DELOU GOTO 1 RHSOU = RARG1 I•ELOU = RARG2 GOTO 1 IF<LARGS,GT,01 GOTO 181 WRITE ILWD, 11901 OIUITJ IRHSOU)l,DELOUJ - SET OUIT CRITERIA C I 191 C C 181 180 C C 1181 C I = IARG 2 - IARG1 + 1 NOPTS = NOPTS - I XHAX = XHIN + INOPTS-11*XDEL IFIIARGl,GT,NO PTSJ GOTO 1 IYIIARG11 = IY<IARG1+I J IARGl = IARG1 + 1 GOTO 163 WRITEILWD,1130) IARG1,IARG2 WRITEILWD,VFRHATJ IIYIIJ,I=IARG1,IARG2) GOTO 1 K<ILL J I MJl, NJ - DELETE POINTS C C < 16 1 C = IA RG2 = NOP TS IF<IAR G1, LE,O, OR, IARG1,GT , NOPT S , OR,IARG2 , LT,I ARG1 * ,OR,IARG 2,G T , NOPT S I GOTO 590 I F <COMAND,NE,16 1 GOTO 1 70 u,:• IFIIARG1,LE,O,OR,IARG1 , GT , NO CU RVI GOTO 570. IFIIARG2, LT,O,OR,IARG 2 ,GT,3 1 GOTO 6 0 0 IARGl = IIARGl-11*3 + IA RG 2 IF<IARG2,EO,OI GOTO 131 IARG2 IARG1 GOTO 132 C 2ND ARG ~O - FIX ENTIRE CURVE, 131 IARGl = IARG1 + 1 IARG2 = IARG1 + 2 132 K = FIXED IFICOHAND,E0,141 K = UNFIXD DO 133 I =IARG1,IARG2 FIXARR<I> = K 133 CONTINUE GOTO 1 C C 115> POIINTSJ - ADD TO PREVIOUS DATA POINTS 130 ....... 0 C C 122) C 220 GUCESSI YFILTER,ZFILTER - GENERATE PARAMETER GUESSES GECNERATEI NOPTS,XMIN,XHAX,E - GENERATE DATA FROM PARAMS 250 C C 1251 C DEBUG= IARG1 GOTO 1 DEB<UGI I - SET DEBUGGING FLAG IFCLARGS,LE,01 GOTO 241 DECODE<LARGS,2050,CSTRNGCIPTRl,ERR=5001 * NOPTS,XMIN,XHAX,YMIN IFINOPTS,LE,11 NOPTS = MAXPTS GOTO 2 42 241 NOPTS = MAXPTS XMIN = 0 , XMAX = 1, YMIN = 0, C HAKE SURE A VALID OUTPUT FORMAT EXISTS, 2 42 DO 243 1=1,MAXIFR VFRMATCII = IFRHATCIJ 243 CONTINUE C COPY COMMAND INTO TITLE , DO 244 1=44,80 TITLE< II = ' ' 244 CONTINUE DO 245 I=l,MINO(LSTRNG,331 TITLE<It43) = CSTRNGCII 245 CONTINUE RETURN 240 C C 1241 C IFILARGS,LE,01 GOTO 2 32 YFILT = RARG1 ZFILT = RARG2 GOTO 234 C NO VALUES SPECIFIED, USE DEFAULTS, 232 CALL IMXMINIIY,NOPTS,IYMAX,IYMINI YFILT = 0,05 *IYMAX ZFILT = O,OOl*IIYMAX-IYMINI/XDEL 234 RETURN 230 C C (231 C 222 221 YCSCALEI A,B - SCALE Y DATA 212 YHIN = CYHINtRARGll*RARG2 YHAX = IYMAXtRARGl>*RARG2 DO 221 !=2,NOPAR,3 Pill = IP I I>•RARG11*RARG2 CONTINUE DO 222 I=l,NOPTS IYCII = INTI IFLOATIIYIIlltRARGll*RARG2 CONTINUE RATMAX = RATMAX / RARG2 GOTO 1 DO 212 I=l,NOPAR,3 Pill = IPI IJtRARGl l *RARG 2 Pl lt 21 = P I I+ 2 >*RARG 2 CONTINUE RATHAX = RATHAX*RARG2 GOTO 1 1010 1020 1030 1040 C C C 600 590 580 570 560 550 540 530 520 510 50 0 MAIN l 'f,(lG f/AM FOR EXIr *' *' *' *' FORMATC ' tINPUT FILE! ' I FORMATC ' tTITLE! ' ) FORHATC ' fOUTPUT FILE! 'I FORMAT(' GELFIT Command Summary!'/ * ' OThe minimum reouired letters of each command or * oPtion'/ ' are shown, All commands are terminated * bY the return key,'/' <list > indicates a set of * oPtional arguments, ' / * ' OA <list >' ,T15, ' ADD more curves ' / *' B k',T15,'fit k BANDS wide ' / * ' C' ,T15,'CLOSE outPut file'/ * ' D i,J',T15, ' DELETE curves i-J'/ *' FIT k,l',T15,'FIT kits with 1 cits'/ *' FIX i,J ' ,T15, ' FIX curve i Param J '/ * ' GE i,xl,x2,e',T15,'GENERATE data from Params ' / *' GU yfil,zfil ', T15,'GUESS curve Parameters '/ H' ,T15, ' tYPe this HELP summary'/ *' /L',T15,'tYPe SY!GELFIT,HLP'/ * ' I infile ' ,T15,'get INPUT file'/ /A',T15,'ADD to previous data'/ K ~,n ' ,T15,'KILL Points •-n'/ *' L m,n',T15, ' LIST Points m-n' / *' H i,J ' ,T15,'MODIFY curves i -J'/ * ' N',T 15,'get NEW curves '/ *' 0 outfil ' ,T15,'set OUTPUT filename'/ /LI ' ,T20,'set LIST filename '/ *' /PL',T20,'set PLOT filename '/ OUTPUT FORMATS CALL ERRORC ' F ' , ' illegal command - tYPe H<cr ) for helP ,') GOTO 1 CALL ERRORC ' F ' , ' ambiguous co mmand,') GOTO 1 CALL ERR ORC ' F ' ,'illegal oPtion,'I GOTO 1 CALL ERRORC ' F', ' ambiguous oPtion, ' I GOTO 1 CALL ERRORC'F','no Points, ' ) GOTO 1 CALL ERROR C' F', ' no curves,') GOTO 1 CALL ERRORC ' F ' , ' no more curves,'> GOTO 1 CALL ERRORC ' F ' , ' bad curve number,') GOTO 1 CALL ERRORC'F ' ,'no more Points,') GOTO 1 CALL ERRORC'F ' , ' bad Point number,') GOTO 1 CALL ERRORC ' F', ' bad Parameter number,') GOTO 1 ERROR MES SAGE S r,ETUF,N ro C C C - RE TUF, N GOTO l [(XIT) 260 C C L'b> C t-' -..J •••N • •••••X••• .••• • •••••y••••••••••••H• • ••••••'> FORMAT<2I61 FORMAT(3F12,0) FORMAT<80A1) FORMAT<4F8,0,314) FORMAT<I4,3F8,0) END SUBROUTINE SETFIL INPUT FORMATS FORMATCI5,31G12,5,A1,1XII FORMATl'SNEW VALUES! 'I FORMAT( ' SXMIN,XINC! ') FORMAT< ' ENTER DATA, ONE INTEGER PER LINE 1-z=ENDl ' I FORMAT( ' ODATA POINTS ' ,14, ' TO ' ,14, ' ! ' 1 FORMAT('SMAX,REL,HALFWIDTH! ' I FORMAT('SRMS OUIT! ' ) FORMAT( ' SMAX,CHANGE OUIT! 'I FORMATC ' SFACTORS! ADDITIVE, MULTIPLICATIVE! ' I FORMATC'SLISTING FILE! ') FORMAT( ' SPLOT FILE! ' I FORMAT( ' t • Y a,b',T15,'YSCALE (s+al*b ' /1 FORMAT( ' SSTANDARDS FILE! ' ) FORMATC'OBND MAT TOO LRG' /' tP=',16,' tN= ' ,I6, *' MAXBND= ' ,16/1 FORMAT( ' SENTER CU RVES - X,Y,H <CR > c-z=ENDJ: ' ) PLTFIL I I I CONlINU E REf URN END LSTFIL I II 31 29 30 28 =0 LSTFILC1): 'L' LSTFIL<21 : 'P' LSTFILC31 : ' I ' DO 29 J=1,6 IFCINPFIL<I+Jl,EO,',') GOTO 30 LSTFIL<J+3> : INPFILCI+J> CONTINUE DO 31 I=J+3,16 LSTFIL<I> CONTINUE DO 32 I:1,16 I =I +1 GOTO 5 C FIND COMMAND NAME IN COMSTR ARRAY, 6 IPTR: 1 COMNUM: COMIND<CSTRNG,COMSTR,COMPTR,MAXCOM,iPTRi IF< COMNUM,LE , O) GOTO 50 C VALID COMMAND FOUND, CHECK FOR SWITCHES, C IPTR HAS BEEN SET TO THE FIRST CHAR AFTER THE COMMAND NAME, I IFCLSTRNG,GT,O> GOTO 3 CALL GTLIN<CSTRNG,PROMPT> DO 1 LSTRNG: 1,81 IF<CSTRNG<LSTRNG>,E0,01 GOTO 2 CONTINUE LSTRNG: LSTRNG - 1 2 GOTO 4 CSTRNG(LSTRNG+1>: 0 3 C CLEAR SWITCH FLAGS, I : 1 4 IF<I,GT,MAXFLG) GOTO 6 5 SWIFLGC3,JI : 0 /10, ,,, , 128/ COMMON/FILCOM/OUTFIL<161,INPFIL<16),LSTFIL<161, *PLTFIL<161 BYTE OUTFIL,INPFIL,LSTFIL,PLTFIL C DEFAULT PLOT AND LIST FILES ARE LP!XXXXXX C FIND END OF DEVICE NAME, IF SPECIFIED DO 27 J:1,4 IFCINPFILCil,EO,•:•J GOTO 28 27 CONTINUE DATA PROMPt BYTE PROMPT<3) BYTE CSTRNGCLSTRNGl,COMSTR<1l,SWISTR<1> INTEGER LSTRNG,COMPTR<2,MAXCOMl,MAXCOM,SWIPTR(2,MAXSWII, *MAXSWI,SWIFLG<3,MAXFLG>,MAXFLG,IPTR INTEGER I,J,SWINUM INTEGER COMIND C C C C C V00-04 C C R,F,MURPHY 07-NOV-78 C C CONVERTS A COMMAND STRING AND ASSOCIATED SWITCHES TO A C COMMAND NUMBER AND SET FLAGS IN THE SWITCH FLAG ARRAY, C C INPUT ARRAY FORMATS C C CSTRNGCLSTRNGI INPUT COMMAND STRING C COMSTR<> CONSECUTIVE ASCII NAMES OF COMMANDS, C COMPTR<2,11 FIRST ELEMENT IS POINTER TO END OF C FULL COMMAND NAME, SECOND ELEMENT IS C F'OJNTER TO END OF UNIQUE COMMAND NAME, C SWISTR<> SIMILAR TO COMSTR FOR SWITCHES, C SWJPTRC2,1> SIMILAR TO COMPTR FOR SWITCHES, C SWIFLG(3,1) (1,NI IS SWITCH t, C <2,NI IS t OF COMMAND IT IS ALLOWED ON, C <3,NI JS SET TO t OF TIMES SEEN, INTEGER FUNCTION COMNUM<CSTRNG,LSTRNG,COMSTR,COMPTR, *MAXCOM,SWISTR, SWJPTR,MAXSWJ, SWIFLG,MAXFLG,IPTRI C COMNUM,FOR 1~ C C CREATt DEFAULT PLOT AND LIST FIL£NAM£S C 2 000 2010 2020 2040 2050 C C C 1070 1000 1090 1100 1110 11 20 1130 1140 1190 1200 1230 1240 1250 1050 1060 *' *' W a ' ,T15, ' set ma x relative WIDTH'/ *' X a,b',T15, ' XSCALE <x+a l tb'/ * ' PL <list > ' ,T15, ' PLOT da t a a nd f i t ' / *' / C ' ,T15, ' Plot in CRT format ' / *' /LP ' ,T15, ' Plot in LP f o r ~at '/ * ' PO ' ,T15, ' add ~ore POINT S'/ * ' 0 a,b',T15,'set OUIT criteria'/ t• S stdfil ' ,T15, ' ~et STANDARDS '/ *' T i,J ' ,TlS, ' TYPE curves i - J ' / U i,J',T15, ' UNFIX curve i Para~ J ' / N ....... BYTE STRING<1>,STRNGS<1> INTEGER STRPTR<2,HAXINDJ,HAXIND,IPTR ► J = J GOTO 10 20 CONTINUE C NO HATCH, ~5 COHIND = 0 + 1 I = I DO 20 COHIND = 1,HAXIND I= IPTR J =1 C BEGINNING OF STRING N IS 1 PLUS END OF STRING N-1, IF<COHIND,GT,1> J • 1 + STRPTR<l,COHIND- 1> K = SfRPTRCl,COHIND> 10 IF C~iTRING< I> ,GT, • Z ' ,OR,STRINGC I> ,LT, ' A' > GOTO 3-0 I F( J,Gf,K l GO TO 2 5 IF<STRING ( I l ,NE, STRNGS<J> > GOTO 20 C C C IN A LARGER STRING ARRAY AND RETURNS THE STRING NUMBER SEE COHNUH,FOR FOR ARGUHENT FORHAT, C C C FINDS A STRING 07-NOV-78 *HAXIND,IPTR> C R,F,HURPHY C C V00-03 c-----------------------------------------------------------c COHIND,FOR c-----------------------------------------------------------INTEGER FUNCTION COHIND(STRING,STRNGS,STRPTR, C BAD SWITCH 30 COHNUH = -2 GOTO 50 C AHBIGUOUS SWITCH, 35 COHNUH = -3 GOTO 50 C ADVANCE POINTER TO START OF ARGUHENTS, IF ANY, 40 IF<CSTRNG(IPTRl,NE,' ') GOTO 50 IPTR = IPTR + 1 GOTO 40 50 RETURN END C 10 IFCCSTRNG<IPTRl,NE,'/') GOTO 40 C HAKE SURE SWITCHES ARE ALLOWED, IFCHAXSWI,LE,O> GOTO 30 C SKIP THE/, IPTR • IPTR + 1 SWINUH = COHINDCCSTRNG,SWISTR,SWIPTR,HAXSWI,IPTR) IF<SWINUH> 35,30,15 C VALID SWITCH FOUND, SET FLAG AND CHECK FOR HORE, 15 DO 20 J=l,HAXFLG IF<SWIFLG<1,JJ,NE,SWINUH,OR,SWIFLGC2,Jl,NE,COHNUH> * GOTO 20 SWIFLG(3,J) = SWIFLG<3,J> + 1 GOTO 10 20 CONTINUE C NOTHING SHOULD PASS THROUGH THIS [IQ LOOP, REAL HIN,HAX,PLTHIN,PLTHAX INTEGER POWER REAL T,AT SUBROUTINE GUESSG<IY,NOPTS,XHIN,XDEL,IYFILT, flrILT, P ,NOPAR,HAXPAR) C V0 0 · 04 28- J llN- / 9 C C R,F,MURPHY CREATED 29 -UE C- 78 C C MAK ES PARAMET ER GUESSES FOR GEL FIT, A CIIRVE I S CREAT ED C BUES SG, FOR T = AHAX1CABSCHINJ,ABSCHAX>J POWER• 0 IFCT,EQ,01 GOTO 10 C ALLOWABLE LIHITS ARE A,B*lO**LOGCX>, C WHERE A AND BARE SINGLE DIGITS, T = ALOGlOCT> AT= AINT<T> POWER= INTCAT) - 1 C ROUND POWER [10WN FOR NEGATIVE LOGS, IF<AT,GT,T) POWER• POWER - 1 C ROUND DOWN FOR HINIHUH, 10 T = HIN/10,**POWER AT = AINTCT> IFCAT,GT,T) AT= AT -1, PLTHIN = AT*10,**POWER C ROUND UP FOR HAXIHUH, T = HAX/10,**POWER AT = AINTCT> JFCAT,LT,T> AT= AT+ 1, PLTMAX = AT*lO,**POWER RETURN END C C C CALCULATES DtFAULTPLOT LIHITS rROH ACTUAL C DATA HIN AND HAX, C CREATED 27-DEC-78 SUBROUTINE PLTLIHCHIN,HAX,PLTHIN,PLTHAXJ 01-FEB-79 C R,F,HURPHY C C V00-01 C C PLTLIH,FOR C GOTO 40 C HATCH SO FAR, SEE IF ENOUGH CHARS WERE SPECIFIED, 30 IFCJ,GT,STRPTRC2,COHIND>> GOTO 40 C NOT ENOUGH HATCHED, IF NONE HATCHED, COHHAND IS ILLEGAL, IFCJ,E0,1> GOTO 25 C SOME HATCHED, BUT COHHAND IS AMBIGUOUS, COHIND = -1 C SAVE END OF HATCHED STRING, 40 IPTR = I RETURN ENI• w --.J LOGICAL*l ur, LFOUND,RFOUND,INFTST REAL YLEFl, YR IGH T,ZLEFT,ZRIGHT,ZLINF,ZRINF INTEGER NOrr s . IY(NOPTSI, NOPAR, MAXPAR, IYFILT REAL XMIN , XDEL, ZFILT, PIMAXPARI E~ 1 lfnr-u - ir•11;t 11 CREATED 29-HAY-79 INTEGER I,J REAL XT,XDEL,YT C C MAKE SURE ARGUMENTS ARE VALID, IFINOPTS,LE,1,0R,NOPAR,LE,Ol GOTO 50 C SAVE ORIGINAL PARAMETERS, S DO 10 I • l,NOPAR IFIERRP<Il,NE,O,l OLDP<Il • Pill 10 CONTINUE C C GENERATE DATA FROM FUNCTION PARAMETERS, ALLOWING C PARAMETERS TO VARY IN NORMAL DISTRIBUTIONS WITH C SPECIFIED STANDARD DEVIATIONS, DATA IS EVENLY C X-INCREMENTED UNLESS LOGFLG IS ,TRUE,, IT IS CREAL UNLESS INTFLG IS ,TRUE,, AND IT IS C XIY UNLESS YONFLG IS ,TRUE , C OLDP NEED ONLY BE SUPPLIED IF AT LEAST ONE ERRP IS NON-ZERO, CONLY THE NECESSARY SUBSET OF X,Y,IX,IY <DETERMINED BY C THE INTFLG AND YONFLG SETTINGS) NEED BE SUPPLIED, C INTEGER NOPAR, NOPTS LOGICAL LOGFLG,INTFLG,YONFLG INTEGER IX<NOPTSl,IYINOPTSl REAL X1,X2,FUNCTN REAL P<NOPARl,ERRPINOPARl,OLDPINOPARl,XINOPTSl,YINOPTSl C *FUNCTN,NOPTS,X,Y,IX,IY,INTFLG,YONFLGl 29-MAY-79 RIGHT> ,NOT,UP & (ZRIGHT-ZRINFl >ZFILT DO SO 1•2,NOPTS-2 ·; r1 N , c-----------------------------------------------------------c GENERD,FOR c-----------------------------------------------------------SUBROUTINE GENERDIX1,X2,LOGFLG,NOPAR,P,ERRP,OLDP, C C VOl C C R,F,HURPHY YRIGHT • IIYIIl+2,*IYII+1l+IYII+2>ll4, ZRIGHT • IYRIGHT-YLEFTl/XDEL C SEE IF LEFT INF POINT FOUND ALREADY, IFILFOUNDl GOTO 10 CLEFT INF POINT REQUIRES UPI DON'T CHANGE ZLINF UNTIL TRUE, IFl,NOT,UPl GOTO 40 IFIZLEFT,GT,ZLINFl ZLINF • ZLEFT IF( <ZRIGHT-ZLINFl ,GE,ZFILTI GOTO 40 C FOUND LEFT INF POINT, INIT POINTERS, LFOUND • , TRUE, • RFOUND • ,FALSE, ZRINF • ZRIGHT !LEFT• 1-1 IMAXP • I GOTO 40 ·c ALREADY FOUND LEFT, UPDATE MAX POINTER, IFIIYll),GT,IY(IMAXPll IMAXP • I 10 C CHECK RIGHT INF CRITERIA, INFTST • 1 , NOT,UP,AND,IZRIGHT-ZLEFTl,GT,ZFILTl C : 1: </f \ .J 1 ,H I. L.J . , .h· .1N1 I l)W /' :·:, :J EE JF f''R[ 1) 1UU :., i Ul NT l'h::;:-;Eu ,_ RJ lfldA, l.f' \ f, I 111.JN [.I J GOHJ 20 f, FOUNI • " !N f' lST GOTO 40 C LASr f''()l'II l•'AS:·;u,. IF THIS DD ES r,[_('.(l , CO NTINU E SEARCH, 20 IF\INFISTJ GOTO 40 C REGIO N PASSES/ CRITERIA, CHECK Y FILTER, IFIIY<IMAXPJ,LE,IYFILT> GOTO JO J ~ NOPAR + 3 I FI J, GT,MAXPARl GOTO 60 NOPAR • J PINOPAR-2l • XMIN t IIMAXP- ll*XDEL P <NOPAR - 1) • FLOATIIY<IMAXPJl P(NOPARI • 11-ILEFTl*XDEL / 4, LFOLIND • ,FALSE, 30 40 UP• <ZRIGHT,GT,ZLEFT> YLEFT • YRIGHT ZLEFT • ZRIGHT 50 CONTINUE GOTO 70 60 CALL ERROR('W','Para~ arra~ full - curve(sl lost,') 70 RETURN L: I IF'!'IATF LEFT> UP & IZRIGHT-ZLINFl <ZFILT C CURVE CRITERIA! C IYIIMAXPl >IYFILT, C WHERE IMAXP IS BETWEEN INFLECTION POINTS, C C C C C C 'UP' IS A FLAG FOR THE PREVIOUS CONDITION ZRIGHT >ZLEFT C INFLECTION POINT CRITERIA! C C START CHECK LOOP, C C SMOOTH THEY DATA USING A SIMPLE r OLYNOMIAL SMOOTH, C SET UP INITIAL VA !.UES, NOPAR • 0 UP• ,FALSE , LFOUND • ,FALSE, YLEFT • 13,tlY111+IYl21l/4, YRIGHT • IJY<ll+2,*IYl2l+IYl3ll/4, ZLEFT • IYRI GHT-YLEFTl/XDEL C REOUIRE INFLECTION POINT TO BE GREATER THAN LEFT BOUNDARY, ZLINF • ZLEFT YLEFT • YRIGHT C C C C PETWEEN LOCAL MAX I MA AN D MINI MA or THE C FIRST DERIVATIVE \ INFL EC:I !ON l'OINJ'.; ) , C THE SE MAXIMA fl NL1 M[NlMA MU S l l<f. Uf,LAIE.I , 111,~ N C THE Z FILTER, •1 N[1 !HF r MA ~I MLIM ! N C THAT INTERVAL MUST fl!" GREATER ! HAN lflE Y f [l! U,, C REASONABLE VALUE S FOR lHE FILlERS AR[ ! C IYFILT = 0,05*1YMAX C ZFILT • 0 ,00l* lli MAX - IYMINI /X DEL -.J ~ C C OUTPUTS C c-----------------------------------------------------------c FTGAUS,FOR SUBROUTINE FTGAUSIBANDPP> C V04-04 28-JUN - 79 C C R,F,MURPHY CREATED 1-JUL-77 C C THIS SUBROUTINE FITS GAUSSIAN COMPONENTS USING A C MODIFIED DAMP LEAST SOUARES ALGORITHM DESIGNED BY HAROUARDT, CJ, SOC, IND~ APPL, HATH, (1965) 11l431-441, C RETURNS CHISO AND DEGREES OF FREEDOM C FOR SIGNIFICANCE CALCULATION, C C INPUTS C C LWD /LOGDEV/ LOGICAL OUTPUT DEVICE C XMIN /DATA/ X VALUE ASSOC, WITH IYll) C XMAX /DATA/ X VALUE ASSOC, WITH IY I NOPTS ) C NOPTS /DATA/ t OF DATA POINTS C IYCNOPTS> /DATA/ ARRAY OF Y DATA VALUES C NOCURV /C URVES/ NOPAR/3 C NOPAR /CURVES/ t OF FIT PARAMETERS C PINOPAR) /CURVES/ PARAMETER ARRAY C NIT /FITPAR/ t OF ITERATIONS TO TRY c------------------------------------------------------------ C [l[[IUG NClT RMi; ou [1[UlU F(A fMAX NFPAR X[1[L FIXED FI XARR I NOF'AR) GELFUN / FI !PAR / / FI lTAf' / / F I lf 'AR / /F ITf'AFU / F IlF'AR / / FITPAR / / FIXCOM/ / FIXCOM / <GLOI<AL > /DEBUG/ t OF CORRECTIVE TRil S f'ER ITERATION OUIT WHEN RMS REAC HE S THIS VALUE OUIT WHEN MA X CHANGE REACHES THIS MAX RATIO OF HW TO YMAX t PARAM WIDTH OF I<AND MATRIX CHANGE IN X I<ETWEEN IY(l ) & IYII+l ) CHARACTER FOR FI XE D PARAH INDICATES WHICH PARAMS TO FIX CALCULATES GAUS SIAN FUNCTION DEBUGGING FLAG C C C C 5. C 4, C 3, C C C 2. /FITPAR/ /FITPAR/ /CURVES/ /FITPAR/ / FITPAR/ NEW PARAMETER VALUES • or [IEGREES OF FREEDOM CONDITION t WHICH CAUSED QUIT 0 = NIT DONE 1 = NCIT EXCEEDED 2 = DELQU ATTAINED 3 = RMSQU ATTAINED 4 = EQUATIONS SINGULAR 5 ~ RATHAX FAILURE ROOT HEAN SOUARE ERROR OF FIT 'GOODNESS OF FIT' DECOMPOSE BANDPP MATRIX SOLVE DECOMPOSITION FOR CHANGE VECTOR, STORE PARAMETERS, CALCULATE PARTDX,PARTDY,BANDPP,DIFFSO, CALCULATE RMS, CHISO, OUTPUT RMS, CHISO IF ANY ITS DONE, CHECK TERMINATION, IF DONE, RETURN, SCALE PARTDY AND BANDPP, ADD CORRECTION FACTOR (INITIALLY 0,01) TO COLUMN NFPAR OF BANDPP, DECOMPOSE BANDPP AND SOLVE FOR DELTA, METHOD: C C C 1, RMS CHISO PCNOPAR> NFREE OCONI1 C C C C C C C C C C C IBAN[1T BANSOL Ul C C --.J C MODULES USED PARCAL DELMX PARTDXINOPAR) PARTDYCNDIFF> DELTACNDIFF> OLDP(NOPAR) IHAPCNDIFF) <ARG> PARTIAL OF EACH PARAH TO EACH PARAM IN LOWER HALFi SPACE FOR DECOMPOSITION OF BANDPP IN UPPER HALF /PARCAL/ GELFUN FLAG TO CALCULATE PARTDX /FI Tf•AR/ MAX REL CHANGE IN ANY PARAMETER /PARTDX/ PARTIAL DERIV WITH RESPECT TO X /PARTDY/ SCALED PARTIAL DERIVATIVES /DELTA/ AMOUNT TO CHANGE PARAMS BY /OLDP/ STORAGE FOR PREVIOUS PARAHS /IMAP/ MAPS ORIGINAL TO VARIED PARAHt'S BANDPPCNDIFF*NFPAR*2> C C C C C C C C C C C C C C C WORKING ARRAYS AND VARIABLES <COMMONS ARE DIMENSIONED HERE FOR 30 CURVES C SO THAT THEY ARE INCLUDED IN WHICHEVER C OVERLAY THIS MODULE IS PLACED !Ni C COMMONS HAY BE EXTENDED BY MAIN PROGRAM C IN WHICH CASE THEY WILL BE IN THE ROOT SEGMENT) C C C C C C C C C C C XDEL = 0, IF<NOPT S ,GT , 1) XDEL = IX2 - Xl )/I NOrrS - l ) XT = Xl IFl,NOT,LOGFLG> GOTO 15 XDEL = 10,**XDEL XT = 10,**XT C CALCULATE NEW PARAMETERS FOR EACH POINT, C AND EVALUAT E THE FUNCTION, 15 DO 30 I =l ,NOPTS DO 20 J =l ,NOPAR IF<ERRPCJl,EO,O,l GOTO 20 PIJ) = GAUSIOLDPIJl,ERRPIJ)) 20 CONTINUE IFIYONFLG> GOTO 24 IF<INTFLG> GOTO 22 XII) = XT GOTO 24 22 I XII) = INTCXT> 24 YT= FUNCTN IXT> IFIINTFLG) GOTO 26 YII> = YT GOTO 28 26 IY<I> = INTIYT) 28 IFILOGFLG> GOTO 29 XT = XT + XDEL GOTO 30 29 XT = XUXDEL 30 CONTINUE C RESTORE ORIGINAL PARAMETERS, DO 40 1=1,NOPAR IFIERRP<Il,NE,0,) PCI> OLDP<I> 40 CONTINUE 50 RETURN END REAL BAN[IPP( 1) COMMON/LOGDEV/LRD,LWD INTEGER LRD,LWD COMMON/DAfA/TITLE<BO>,XMIN,XMAX,YMIN,YMAX, *NOPTS,MAXPTS,IY<l> BYTE TITLE INTEGER NOPTS,MAXPTS,IY REAL XMIN,XMAX,YMIN,YMAX COMMON/CURVES/NOCURV,NOPAR,MAXPAR,P<l> INTEGER NOCURV,NOPAR,MAXPAR REAL P COMMON/FITPARINIT,NCIT,RMSOU,DELOU,RATMAX,NEIGHB,NFPAR, * MAXBND,XDEL,NFREE,OCOND,RMS,CHISO,DELMX INTEGER NIT,NCIT,NEIGHB,NFPAR,MAXBND,NFREE,OCOND REAL RMSOU,DELOU,RATMAX,XDEL,RMS,CHJSO,DELMX COMMON/PARCAL/PARCAL LOGICAL PARCAL COMMON/FIXCOM/FIXED,UNFIXD,FIXARR<t> BYTE FIXED,UNFIXD,FIXARR COMMON/DEBUG/DEBUG INTEGER DEBUG COMMON/PARTDX/PARTDX<90) REAL PARTDX COMMON/PARTDY/PARTDY(90) REAL PARTDY COHMON/DELTA/DELTA<90) REAL I•EL TA COMMON/OLDP/OLDP(90) REAL OLDP COHMON/IMAP/IMAP(90) INTEGER IMAP SCALE DELTA, ADD XKGAM*DELTA TOP, CALCULATE DJFFSO, IF FIT WORSE: CHECK COSINE CRITERION! GT, DECREASE XKGAM, GOTO 6, LE, INCREASE FACTOR, GOTO 5, FIT BETTER, CHECK HALFWIDTHS BAD, CORRECT AND GOTO 2, FIT OK, DECREASE CORRECTION FACTOR, GOTO 1, REAL FUNCTN EXTERNAL FUNCTN LOGICAL*! ALLOK C ITC IS THE CORRECTION fTERATION COUNTER, C ITNO CONTAINS THE CURRENT ITERATION NUMBER, C I-- >M ARE TEMPORARY VARIABLES, INTEGER I,IPMAX,ITC,ITNO,J,K,L,H - C THES£ VARIABLES HAVE SPECIFIC VALUES DURING THE FITTING, INTEGER NDIFF,NDIFFl,NFPARl,NPPSIZ,NPPSZl C MISCELLANEOUS VARIABLES, INTEGER IYFUNV,IYDATV REAL DIFFN,DIFFSO,DIFF,DIFF2,FLAM,XKGAM,XLAH, *GAMNUH,TEHP,TEMPSO,DEL,DELSO,STIME,X C PROCEDURE TYPE DECLARATIONS, INTEGER !BANDT REAL GELFUN C*** C C C C 6, C C C C C C C C RU1L SEC: Nf!S C IF PARCAL=,TRUE,, 'GELFUN' WILL CALCULATE PARTDX, PARCAL = <ITNO,LT,NIT,AND,ITC,LE,NCIT) 15 IF(,NOT,PARCALJ GOTO 35 DO 20 I=l,NDIFF PART[IY( I) = ◊-, CONTINUE 20 DO 30 I=l,NPPSIZ BANDPP ( IJ = 0, 30 CONTINUE 35 DIFFSO = 0, CHISO = 0 , X = XMIN C C INITIALIZE VARIABLES, 4 NFREE = NOPTS - NDIFF NDENOH = MAXO<l,NFREEJ IF<NFREE,LE,0,) * CALL ERROR('W ' ,'solution underdeter~ined,') IF(NFPAR,GT,NDIFFJ NFPAR = NDIFF NFPARl = NFPAR - 1 NDIFFl = NDIFF - 1 NPPSIZ = NDIFF*NFPAR NPPSZ1 = NPPSIZ t 1 XLAM = 0,01 OCOND = 0 ITC= 0 ITNO = 0 ALLOK = ,TRUE, IF<NIT,GT,Ol WRITE<LWD,1000) CHAIN ITERATION LOOP, C SAVE PREVIOUS PARAMETER VALUES, DO 10 I=1,NOPAR 5 OLDP(IJ = P<IJ 10 CONTINUE C RETURN HERE IF RATMAX FAILS, C CALCULATE PARTIAL DERIVATIVE MATRICES, C BANDPP IS ACTUALLY BANDPP<NDIFF,NFPAR>, WHERE BANDPP<I,J> C IS THE PRODUCT OF THE PARTIALS OF PARAHS I AND I+NFPAR-J, C THUS, THE SOUARES OF THE PARTIALS OF THE PARAMS ARE IN C COLUMN NFPAR, [1 GOTO 1 IF ( [1Ef<UG, NE, 0 > * CALL ERROF,( ' W' , ' [1et,1_,si,Hn<l code not con,r-iled, ' ) IF<I,GT,NOPAR) GOTO 3 IF<FIXARR(I):Eo,FIXED) GOTO 2 NDIFF = NDIFF t 1 IMAP<NDIFFJ = I 2 I=I+l GOTO 1 3 IF<NDIFF,GT,OJ GOTO 4 CALL PARAMETERS ARE FIXED, DON'T DO ANY ITERATIONS, NIT= 0 NDIFF = 1 IMAP(ll = 1 I = 1 C CREATE MAP ARRAY TO POINT ro UNFIXED PARAMETERS, NDIF F = 0 C --.J Q\ 66 64 63 c•••- ITNO = ITNO + 1 IFIITNO,LE,NITl GOTO 63 GOTO 400 IFIITC,LE,NCIT) GOTO 64 OCOND = 1 IFl , NOT,ALLOKl OCOND = 5 GOTO 400 IF<DELMX,GT,DELOU,OR,ITNO,E0,1) GOTO 66 OCOND = 2 GOTO 400 FIRMS,GT,RMSOU) GOTO 68 QCOND = 3 DO 60 I=l,NOPTS C ROUND OFF FUNCTION VALUE TO ACCURACY OF DATA; C MINIMIZES EFFECT OF SMALL DIFFERENCES ON CHISO AND RMS, IYFUNV = INT<GELFUN(Xlt, 5 1 IYDATV = IY<II DIFF = IYDATV - IYFUNV DIFF2 = DIFF*DIFF DIFFSO = DIFFSO + DIFF2 C MINIMIZE ERROR DUE TO POINTS WHERE YCXI = 0, IF CIYFUNV,E0,01 IYFUNV = IYDATV C DIFF2 = 0, IF IYFUNV = 0 , IF <IYFUNV,NE,O> CHISO = CHISO + DIFF2/IYFUNV X = X + XDEL IF<,NOT,PARCAL) GOTO 60 DO 50 J=l,NDIFF PARTDY<J> = PARTDY<J) + DIFF*PARTDXIIMAP(J)I K = NPPS IZ - NDIFF + J L = MINO<J+NFPARl,NDIFF) DO 40 H• J,L BANDPP(KI= * BANDPP(Kl+PARTDX(IMAP(Jl l *PARTDX<IMAP(M)) K = K - NDIFF1 40 CONTIN UE 50 CONTINUE 60 CONTINUE D IF(DEBUG,EO,Ol GOTO 59 D CALL REMARK('AFTER 60 ' ) D CALL REMARK< ' PARTDY'l D WRITE<LWD,3000) (PARTDY<I> , I=1,NDIFF) D CALL RCTRLO D CALL REHARK( ' BANDPP 'l D WRITE<LWD,3000) <BANDPPIIl,I=1,NPPSIZl D CALL RCTRLO 059 CONTINUE C CALCULATE ERROR VALUES AND CHECK TERMINATION CONDITIONS, RMS= SORTIDIFFSO/NOPTSl I F IITNO,EO,Ol GOTO 62 C***+ STIME = SECNDSISTIMEl IF<ITC,GT,NCIT,OR,,NOT,ALLOK) GOTO 61 WRITEILWD,1010) STIME,RMS,DELMX,IPMAX,PIIPMAXl,ITC C WRITEILWD,1010) RMS,DELMX,IPMAX,P<IPMAXl,ITC ITC= 0 GOTO 62 61 WRITEILWD,1010) STIME C REMEMBER TIME EACH ITERATION IS STARTED, 62 STIME = SECNDSIO,l GOTO 400 C FILL rARTDX WITH THE ABSOLUTE VALUE OF THE PARTIAL OF C EAC H UNFIXED PARAM, SCALE MATRI CES BY PARTDX , 68 DO 80 I=l,NDIFF K = NPP SIZ - NDIFF + I PARTDX(II = SORT <BANDPP<K >> IF <PARTDX(Il,NE,0,1 PARTDY<I> = PARTDY(Il / PARTDX<I ) L = MAXO<l,I - NFPARl) DO 70 J=I,L, - 1 X = PARTDX<Il*PARTDXIJ) IFCX,NE,0, ) BANDPP(K ) = BANDPPIKI / X K = K - NDIFF 70 CONTINUE 80 CONTINUE D IFIDEBUG,EO,Ol GOTO 82 D CALL REMARK( ' AFTER 80') D CALL REMARK< ' PARTDX'l D WRITE<LWD,3000) <PARTDX<I>,I=1,NDIFF) D CALL RCTRLO D CALL REMARK('BANDPP'l D WRITE(LWD,3000) (9ANDPPCil,I=1,NPPSIZ) D CALL RCTRLO 082 CONTINUE C INITIALIZE CORRECTION VARIABLES, FLAM= 0, XKGAM = 1, C CORRECTIVE LOOP, C ADD FACTOR TO COLUMN NFPAR, 100 FLAM= XLAM - FLAM DO 110 I=NPPSIZ-NDIFF1,NPPSIZ BANDPP<Il =FLAM+ BANDPPIIl 110 CONTINUE C DECOMPOSE BANDPP, IF<IBANDT<BANDPP,BANDPPINPPSZll, * NDIFF,NFPARl,NE,Ol GOTO 130 C EOUATIONS ARE SINGULAR, RESTORE OLD PARAMS AND FORCE RETURN, OCOND = 4 DO 120 I=l,NOPAR P ( I ) = OUIF' <I > 120 CONTINUE ITNO = NIT + 1 GOTO 5 C SOLVE FOR DELTA (PARAMETER CHANGE VECTOR>, 130 CALL BANSOL<BANDPPINPPSZ1l,NDIFF,NFPAR,DELTA,PARTDY) D IF<DEBUG,EO,Ol GOTO 132 D CALL REMARKl'AFTER 130 ' ) D CALL REMARKl'PARTDX'l D WRITEILWD,3000> <PARTDX(Il,I=l,NDIFFl D CALL RCTRLO D CALL REMARK< ' DELTA'l D WRITEILWD,3000) IDELTAIIl,I=l,NDIFFl D CALL RCTRLO D132 CONTINUE C SCALE THE DELTAS, DO 140 I=l,NDIFF IF<PARTDXIIl,NE,O,l DELTA<Il=DELTAII)/PARTDXIIl 140 CONTINUE C CORRECTIVE LOOP FOR FITS THAT MEET COSINE CRITERION, DO 210 I=l,NDIFF 200 PIIMAP<Ill = P<IMAPII)l + XKGAM*DELTA<Il ...... ...... CONTINUE IF<DEBUG,E0,01 GOTO 212 CALL REMARK( ' AFTER 210 ') CALL REMARK( ' P ') WRITE<LWD,30001 (P(l),l=l,NOPAR) [t CALL RCTRLO D212 CONTINUE C CALCULATE ERROR OF NEW FIT, DIFFN = 0, PARCAL = ,FALSE, X = XMIN DO 2 20 1= 1,NOPTS DIFF = IY<I> - INT(GELFUNCXl+,51 DIFFN = DIFFN + DIFF*DIFF X = i + XDEL 220 CONTINUE C IF FIT IS BETTER, JUMP OUT OF CORRECTIVE LOOP, IF<DIFFN,LT,DIFFSO> GOTO 300 C FIT IS WORSE, RESTORE OLD PARAMS AND CHECK COSINE CRITERION, [t IF<DEBUG,NE,OJ CALL REMARKC ' FIT WORSE ' ) DO 230 I=l,NOPAR PCI) = OUIP(I) 230 CONTINUE ITC = ITC + 1 IF<ITC,GT,NCIT) GOTO 5 C CALCULATE VALUES FOR COSINE CORRECTION, GAMNUM = 0, TEMPSO = 0, DELSO = 0, DO 240 I=l,NDIFF TEMP= PARTDYCI) DEL= [IELTA(I) GAMNUM = GAMNUM + TEMP•t•EL TEMPSO = TEMPSO t TEMP*TEMP DELSO = DELSO + DEL*DEL 240 CONTINUE C THAT'S COSCPI/4), TEMP= SORTCTEMPSO*DELSO) * 0,7071067812 IF(GAMNUM,LE,TEMP> GOTO 250 C COSINE CRITERION MET, DECREASE AMOUNT ADDE[I TO PARAMS, XKGAM = XKGAM/2, GOTO 200 C CRITERION NOT MET, INCREASE AMOUNT A[IDED TO BANDPP, 250 XLAM = 10,*XLAM GOTO 100 C FIT IS BETTER, CHECK HALFWIDTH RATIOS, IFCRATMAX,EO,O,> GOTO 308 300 [I IFCDEBUG,NE,0) • [I * CALL REMARK('FIT BETTER - CHECKING RATMAX') ALLOK = ,TRUE, DO 302 I=3,NOPAR,3 TEMP= P(I-1> IFCTEMP,E0,0,1 GOTO 302 IFCABS<P(II/TEMP>,LE,RATMAX) GOTO 302 ALLOK = ,FALSE, C RATMAX FAILURE, NARROW FAILING CURVE TO YMAX*RATMAX*,8 , C WIDEN PRECEDING CURVE, EITHER HW* 2 OR YMAX*RATMAX*,8, PCI) = RATMAX*TEMP*,8 IFCI,GT,3) P(I-3> =AMIN1CP<I-3 >*2,,P(I-41*RATMAX*,81 . 302 CONTINUE 210 D D D D FORMATC911X,G12,5)) END FORMATC//1X, ' £TCSEC)',5X, ' RMS',7X,'MAX,REL,CHANGE IN *' NEW VALUE',4X,'CITt') FORMATC//5X,'RMS',7X,'MAX,REL,CHANGE IN Pt', *' NEW VALUE',4X,'CITt') FORMATC2X,F5,0l2X,2G14,7,' C',I2,')',G14,7,I4) FORMATC1X,2G14,7,' C',I2,'),G14,7,I4> CREATED 17-[IEC-78 C FOR USE WITH GELFIT, C RETURNS VALUE OF FIT AT X, C C INPUTS C C X <ARG > VALUE TO EVALUTE FUNCTION AT /CURVES/ NUMBER OF CURVES IN FIT C NCRVS / CURVES/ FIT PARAMETERS C PC3,NCRVS) /PARCAL/ IF , TRUE, DO PART DX CALCULATION C PARCAL C C OUTPUTS C <RESULT > VALUE OF FUNCTION AT X C GEL FUN C PARTDX<3,NCRVSI /PARTDX/ PARTIAL DERIVATIVE OF EACH PARAH C REAL X C C R,F,MURPHY c C VOO c-----------------------------------------------------------REAL FUNCTION GELFUNCX) c GELFUN,FOR c------------------------------------------------------------ 1010 C1010 C***[13000 C ClOOO 1000 CU- ♦ Pt ' , IF( ALLOKI GO TO 308 CAT LEAST ONE HW TOO BIG, ITC= IT C + 1 GOTO 15 C ACCEPTABLE FIT, DECREASE AMOUNT A[IDE[I TO DIAGONAL OF BAN[IPP, C CALCULATE THE MAXIMUM RELATIVE CHANGE IN A PARAH, C AND GO BACK FOR NEXT ITERATION, 308 XLAM = XLAM/10, D IF<DEBUG,NE,O> CALL REMARK( ' FIT ACCEPTABLE ' ) DELMX = 0, !PMAX = 1 DO 310 I=l,NOPAR DEL= ABS<P<I> - OLDP<I>> IF<OL[IPCI),NE,O,) DEL= ABSCDEL / OLDP(III IF(DEL,LE,DELMX) GOTO 310 DELMX = DEL IPMAX = I 310 CONTINUE GOTO 5 C FITTING TERMINATE[!, C MAKE SURE ALL HALF-WIDTHS ARE POSITIVE, 400 DO 410 I=3,NOPAR,3 P(I) = ABS(PCI)) 410 CONTINUE 500 RETURN 00 " GELFUN = 0, DO 20 J = 1,NCRVS IF(P (3,Jl ,EO, O,I GOTO 10 ARG = (X - P 11,Jl)/P(3,JI XARG = ARG*ARG I F<X ARG,GT , 15, 1 GOTO 10 EXARG = EXP( -X ARGI GELFUN = GELFUN + P(2,Jl*EXARG IF ( ,NOT,PARCALJ GOTO 20 PARTDX<2,J I = EXARG AMP= 2,*P<2,J)/P (3 ,J) DIDSK • AMP*ARG*EXARG PARTDXl1,JI • DIDSK PARTDX<3,JI • ARG*DIDSK GOTO 20 IFl,NOT,PARCAL) GOTO 20 DO 15 I = 1,3 F'ARTDX<I,JI = 0, CONlINUE CONTINUE RETURN END INTEGER I, J . REAL AMP,ARG,DIDSK,EXARG,XARG COMMON / CURVES /N CRV S ,NP ,M AXPAR,Pl3,l I INTEGER NCRVS ,NP,MA XPAR REAL P COMMON/PARCAL/PARCA~ LOGICAL F·ARCAL COMMON/PARTDX / PARTDXl3,1) REA L PARTDX 21-JUN-77 C rio ~•o K=r· . ~; I F IJ ,NE,M ) GOTO 50 IF IY,LE,0,1 GOTO 210 XL<I,J J=l , / DSQRT IYJ GOTO 90 XLI I,JJ =Y * XL(R,M ) R=r,+ 1 CO NTINUE CONTINUE - 1 !BANDT RETURN l[IANDT 0 RETURN END (l•(l t l Y = Y - XLI l ,Kl*XLIR,ll) SOLUTION OF LY=B DIMENSION XLIN,Ml,Xl11,Bl11 INTEGER I,J,K,P,O,S DOUBLE PRECISION Y S=H- 1 C C C Y=A <I,JI IF (F',GT,S> GOTO 30 SOLUTION OF UX=Y DO 50 I=1,N P=1 IF <I,LT,M) P=M-I+1 O•I Y=B<I) IF <S,LT,PI GOTO 30 r,o 20 K=S,P,-1 0=0-1 20 Y=Y-XLII,Kl*X(Q) 30 XCl)•Y*XL<I,HI 50 CONTINUE D WRITE<S,551 <XCil,I=l,N) D 55 FORHAT<1X,BE10,3) C C C C C SOLVES THE EQUATIONS AX=B USING C !BANDT DECOMPOSITION , ACTUALLY C SOLVES LY=B,UX=Y C C MODIFICATION HISTORY: C 22-AUG-76 ADAPT£D FROH WILKINSON AND REINSCH C LINEAR ALGEBRA 1971 SPRINGER VERLAG CONTRIBUTION 1/4 C BY WILLIAM PEARSON C 18-JUN-77 CONVERTED FOR USE WITH GELFIT C C BY R,F,MURPHY C SUBROUTINE BANSOL<XL,N,H,X,B> C V02-05 C C W,R,F'EARSON, R,F,HURF'HY 22-AUG-76 c-----------------------------------------------------------c BANSOL,FOR c------------------------------------------------------------ 210 90 100 200 50 20 30 O=M-J ♦ P REAL A(N,MJ,XL(N,M) INTEGER N,H INTEGER I,J,K,P,R,S DOUBLE PRECISION Y DO 200 I=1,N P=1 IF <I,LT,MI P=M-1 ♦ 1 R=I-M+P DO 100 J=P,M S= J-1 C FORMS LL* DECOMPOSITION OF ·A SYMMETRIC C POSITIVE BAND MATRIX N*M C SEE BANSOL,FOR FOR REFERENCE, C C R,F,MURPHY C C V02-05 INTEGER FUNCTION IBANDT<A,XL,N,H) c------------------------------------------------------------ c-----------------------------------------------------------c IBANDT,FOR 15 20 10 C C C -...J \.0 10 5 2 C C WRITE<LFD,10101 TITLE,NOPTS,NPAR,NCRVS,NFPAR I FI NOST[l , LE,01 GOTO 2 WRITEILFD,10121 ST[IFIL WRITE I LF[I, 1014 I IF<NOSTD,GT,0 1 GOTO 5 WR ITE<LFD, 1016 1 GOTO 10 IJRITE<LFtt, 10181 TOTP = 0, COHHON/LOG[IEV / LR[l,LWD,LF[I INTEGER LR[l,LWD,LFD COHMON/FILCOM / OUTFIL(161,INPFIL1161,LSTFIL(161, *PLTFIL<16),ST[IFIL1161 BYTE OUTFIL , INPFIL,LSTFIL,PLTFIL,STDFIL COHMON/DATA/T I TLE1801,HINHAX<4>,NOPTS,MA XPTS,IY <1> BYTE TITLE INTEGER NOPTS,HAXPTS,IY REAL HINMAX COHHON/CURVES/NCRVS,NPAR,HXPAR,P<3,11 INTEGER NCRVS,NPAR,MXPAR REAL P COMMON/FITPAR/NIT,NCIT,RMSGU,DELGU,RATHAX,NEIGHB,NFPAR, * HAXBND, XDEL,NFREE,GCON[l,RHS,CHISG,[IELHX INTEGER NI T,NCIT,NEIGHB,NF PAR,MAXBN[l,NFREE,GCOND REAL RHSG0,[IELGU,RATMA X,XDEL,RMS,CHISG,DELHX COMMON/FIXCOM/FIXED,UNFIXD,FIXARR(3,11 BYTE FIXED,UNFIXD,FIXARR COHMON / STDCOH/NOST[l,HAXST[l,LOGSTD,STD<l l INTEGER NOSTD,MAXST[I LOGICAL LOGST[I REAL STD COHHON / WRKSPC/I,J,FRACT,TOTP ,TOTD ,YHEA N, * PERROR,P[IIFF,HW,XT,SUH l ll INTEGER I,J REAL FRACT,TOTP,TOT[l,YHEAN,PERROR,PDIFF,HW,XT,SUH C TYPES THE CURRENT FI T AN[I ERROR ESTIMATES FOR GELFIT, C CREATED 13- [IEC-78 SUBROUTINE GEL LST 01 - JU N-7 9 C R,F,tiURPHY C C V02 c GELLST,FOR c--------------------c------------------------------------------------------------- DO 100 I=N , 1, - 1 P~l I F <<N-I l ,LT,M I P= M- NtI Y=X(II G=I IF <S , LT,PI GOT O 80 DO 70 K=S,P, - 1 G=CH1 70 Y=Y - XL <G, Kl*X<G > 80 X<I J=Y*XL <I,MI 100 CONTINUE RETURN EN[I J 5 I = 1 , NCRVS SUM(Jl=2,*P<2 ,I >*P <3 ,I J TOT P = TOTP t SUM I I I 15 CONTINUE FRA CT = 0, [1 0 20 I=l,NCRVS IF ( TOTP,NE,0,1 FRACT=SUM I I I/ TOTP WRITEILF[I, 10201 * I,(P(J,I l ,FIXARR<J,Il, J= l, 3 1, SUHIIl,FRACT IFINOST[l,GT,01 GOTO 17 C NO STANDAR[IS IN USE, PUT OUT END OF LINE, IJRITE{LF[l,10161 GOTO 20 C CALCULATE AN[I OUTPUT MOLECULAR IJEIGHT, 17 MW = 0, XT = 1, [IQ 18 J=l,NOSTD MW= MW t ST[l(Jl*XT XT = XUP(l,IJ 18 CONTINUE IF(LOGSTDI MW= 10,**MW WRITE<LFD,10221 MW 20 CONTINUE C FIND HEAN OF DATA, TOTD = 0, [IQ 30 I=l,NOPTS TOTD = TOTD + FLOAT<IYIIII 30 CONTINUE YHEAN = TOTD/NOPTS C CORRECT TOTAL AREA FOR BOUNDARIES AND X INCREMENT, TOTD = <TOTD - FLOAT< (IY(lltIY(NOPTSll/21 * XDEL PERROR = 100,,RHS/YMEAN P[IIFF • 100,*ABS<<TOTP-TOTDI/TOTDI WRITE(LF[I, 10301 *TOTP,TQT[l,P[IIFF,RHS,CHISG,NFREE,YHEAN,PERROR IFIRATHAX,NE,0,1 WRITE<LFD,10321 RATHAX I F<GCON[l,EG,01 GOTO 200 WRITE I LFD, 1040 I GOTO<ll0,120,130,140,1501 GCOND 110 WRITE<LFD,10501 NCIT GOTO 200 120 WRITE<LFD,10601 DELGU GOTO 200 130 WRITE(LFD,10701 RHSGU GOTO 200 140 WRITE<LFD,10801 GOTO 200 WRITE<LFD , 10901 RATMAX 150 200 RETURN 10 10 FORMAT< ' 1 ' ///1X,80A1,//1X,I5, ' Points',I3,' Para•eters ' , * 13, ' curves, Band •atri x' ,I3, ' para•eters wide'/) 1012 FORMAT<' UsinS MW standards file ',16A1 /I 1014 FORMAT <7X,'X•a x' ,10x ,'Y•a x ',9x , ' HalfW ' , 10x , ' Area', *6X,'Fraction ' fl 1016 FORHAT< ' t',lX/ 1 FORHAT( ' t ' ,4X, ' Hol,Weisht ' /I 1018 FO RHAT<I4 , 3 ( G12, 5 ,A1,1Xl,2G12,5$1 1020 FORHAT('t ', G1 2 ,5/I 1022 FORMAT( ' OArea(f i tl =',G15,8, ' Area<data l =', 1030 * G15,8, ' ¾diff = ' ,F7,3/ ' 0RHS =' ,G14,7,8X, JI() 0 CXI IF<NXPLOT/4,LT,MAXCOR> GOTO 1 CALL ERRORC'F','not enou9h core for Plot,'> RETURN IFCCRTPLT) GOTO 41 WRITECLFD,1000) TITLE GOTO 42 41 WRITECLFD,1005) TITLE C INDICATE JUST DATA PLOT, 42 IFCNOPAR,LE,01 NYPERX = 0 PARCAL = ,FALSE, CALL TTPLOTCLFD,XFUNCT,YFUNCT,NOPTS,GELFUN, *CRTPLT,XPMIN,XPMAX,YPMIN,YPMAX,NXPLOT,NYPLOT, *NYPERX,CORE,4*MAXCORl RE TURN 1000 FORMATC'l'///lX,BOAl//) 1005 FORMAT<'1',80A1/1 END REAL FUNCTION XFUNCT<I> C C LOGICAL*l FPFLAG EXTERNAL XFUNCT,YFUNCT,GELFUN C END YFUNCT = FLOATCIY<I>I RETURN COMMON/DATA/TITLE(80J,MINMAXC41,NOPTS,MAXPTS,IYC11 £<YTE TITLE INTEGER NOPTS,MAXPTS REAL MINMAX REAL FUNCTION YFUNCT(II END XFUNCT = XMIN + 11 ~11*XDEL RETURN COMMON/DATA / TITLE1801,XMIN l<YTE Tl TLE REAL XMIN COMMON/FITPAR/JUNKllll,XDEL INTEGER JUNK REAL XDEL LLD LOGICAL UNIT TO SEND PLOT TO X X DATA FUNCTION Y Y DATA FUNCTION NOPTS t OF DATA POINTS FUNCTN FIT FUNCTION CRTPLT FLAG (,TRUE. IMPLIES USE CONDENSED FORMAT> XMIN,XMAX,YMIN,YMAX FIT PLOT LIMITS NX t OF COLUMNS IN PLOT NY t OF ROWS IN PLOT NYPERX t OF FUNCTION VALUES TO PLOT PER X WRK TEMPORARY WORK SPACE MAXWRK t OF DOUBLE-WORDS OF SPACE IN WRK C BYTE ERR INTEGER I,IX,IY REAL XINC,XINC2,YINC,XT,YT · C C C C C C C C OUTPUTS C WRK<l,1) t OF DATA VALUES OFF PLOT SURFACE C C LOGICAL*! CRTPLT BYTE WRK I NX, NY> INTEGER LLD,NOPTS,NX,NY,NYPERX,MAXWRK REAL X,Y,FUNCTN,XMIN,XMAX,YMIN,YMAX C C C C C C INPUTS C C PLOTS XlY DATA FUNCTIONS AND/OR A FIT FUNCTION ON A TERMINAL, C CREATED 31-0CT-78 *XMIN,XMAX,YMIN,YMAX,NX,NY,NYPERX,WRK,HAXWRKI 28-JUN-79 C R, F,MURPHY C C V03-01 c-----------------------------------------------------------SUBROUTINE TTPLOTCLLD,X,Y,NOPTS,FUNCTN,CRTPLT, c TTPLOT,FOR c------------------------------------------------------------ C C C C C C PLOT DATA AND FIT FOR GELFIT USING TTPLOT, C INTEGER MAXCOR REAL CORE<MAXCOR> COMMON/LOGDEV/LRD,LWD,LFD INTEGER LRD,LWD,LFD COMMON/DATA/TITLE<SOJ,XMIN,XMAX,YMIN,YMAX,NOPTS BYTE TITLE INTEGER NOPTS REAL XMIN,XMAX,YMIN,YMAX COMMON/CURVES/NOCURV,NOPAR INTEGER NOCURV,NOPAR COMMON/PARCAL/PARCAL LOGICAL PARCAL COMMON/PLOTCM/XPMIN,XPMAX,YPMIN,YPMAX,NXPLOT, * NYPLOT,NYPERX,LOGPLT,CRTPLT LOGICAL*l LOGPLT,CRTPLT INTEGER NXPLOT,NYPLOT,NYPERX REAL XPMIN,XPMAX,YPMIN,YPMAX C SUBROUTINE GELPLTICORE,MAXCORI C VOl-01 28-JUN -7 9 C C R,F,MURPHY CREATED 17-DEC-78 c------------------------------------------------------------ c-----------------------------------------------------------c GELPLT,FOR 1032 1040 1050 1060 1070 1080 1090 * ' CHISO =' ,G14, 7 , 7X, ' NDF =' ,I4 / * ' OYmean = ' ,G15,8,5X, ' XEIRMS / Ymean l =' ,F 7 ,3 //1 FORMAT( ' Maximum ratio of HW to YMAX s et to ' ,914 ,7/ 1 FORMATl ' $Fittin9 terminated with ' I FORMATl ' +limit of ' ,14, ' CITs sur~assed ' / / / 1 FORMATl ' +Ma x ,Rel,Chan9e <= ' ,G11,4 /// I FORMATl ' +RMS <= ' ,Gll,4 /// 1 FORMATl ' +eouations singular ' ///) FORMATl ' +ma x iffium HW / Yma x ratio > ' ,G11,4/// I END to-' 00 INTEGER PLTPNT C C DEFINE PLOT POINTER FUNCTION, PLTPNT<A1,A2,A3 ) = INT< 1,5 + <Al-A2)/A3 C C HAKE SURE THERE'S SOMETHING TO PLOT, ERR= 0 IF<NOPTS,LE,O,AND,NYPERX,LE,O> GOTO 990 C CALCULATE X & Y PLOT INCREHENTs · XINC ~ <XHAX-XHINl/(NX-1) IF<NYPERX,GT,01 XINC2 = XINC/NYPERX YINC = CYHAX-YHINl/(NY-1) HAXNY = HAXWRK / NX IF<HAXNY,LE,01 GOTO 990 C PUT OUT TOP X AXIS, CALL XAXIS<LLD,XHIN,XINC,NX,CRTPLT,'T ' I C !NIT FOR PLOT OUTPUT, C GENERATE AS MANY LINES AT ONCE AS POSSIBLE, NYl = 0 NY2 = HINO(NY,HAXNYI C FILL PAGE IHAGE ARRAY WITH BLANKS 4 DO 10 IX=l,NX DO 5 IY=1,NY2-NY1 WRK< IX, IY> 5 CONTINUE 10 CONTINUE C PUT FUNCTION !HAGE INTO ARRAY IF<NYPERX,LE,01 GOTO 22 XT = XHIN DO 20 JX=l,NX DO 15 1=1,NYPERX YT= FUNCTN<XT) XT = XT + XINC2 IY • PLTPNT<YHAX,YT,YINC) IF(IY,LE,NY1,0R,IY,GT,NY2) GOTO 15 IY = IY - NY1 WRK(IX,IY> = '*' 15 CONTINUE 20 CONTINUE C PLOT DATA OVER FUNCTION, ~SING NUMBERS TO C INDICATE MULTIPLE POINTS, 22 IF(NOPTS,LE,O> GOTO 41 DO 40 J:1,NOPTS IX= PLTPNT(X<Il,XHIN,XINC) IF<IX,LE,O,OR,IX,GT,NX) GOTO 38 IY • PLTPNT<YHAX,Y(I>,YINC> IF<IY,LE,NY1,0R,IY,GT,NY21 GOTO 38 IY = IY - NY1 IF<WRK<IX,IYl,EO,'*'•OR,WRK<IX,IYl,EO,' 'l GOTO 30 C POINT IS ALREADY THERE, BUHP COUNTER, IF<WRK<IX,IYl,GE,'0',0R,WRK<IX,IY>,LT,'9') GOTO 24 WRK<IX,IYl = WRK<IX,IY) + 1 GOTO 40 C USE+ FOR HORE THAN TEN SUPERIMPOSED POINTS 24 WRK(IX,IY) = ' +' GOTO 40 C NO POINT YET, START WITH 0, 30 WRK <IX, I Y> = • 0' GOTO 40 38 ERR= ERR+ 1 10 C C 1050 1044 1020 1022 1024 1030 1034 1040 C IF<TOPBOT,EO,'B'I GOTO 30 IF<CRTPLTI GOTO 12 WRITE<LLD, 1000) GOTO 14 LOGICAL*! CRTPLT BYTE TOPBOT INTEGER LLC•,NX REAL XHIN,XINC INTEGER I,IT,J REAL XT FORHAT('+',E12,5,' -'SI FORHAT<'+',A1SI FORHAT('+- ',E12,5/) FORHAT('+',13X,'1 'SI FORHAT('+l'/1 FORHAT('+',E9,2,'-'S) FORHAT('+-'/) FORHAT('+',9X,'l'SI END SUBROUTINE XAXIS<LLD,XHIN,XINC,NX,CRTPLT,TOPBOTI C FORHAT t'S ENDING IN AO START AN OUTPUT LINE, C THOSE ENDING IN A 2 CONTINUE IT, AND THOSE C ENDING IN A 4 END IT, • C C OUTPUT FORHATS C 40 CONTINUE C PUT OUT PAGE !HAGE ANDY AXES 41 YT= YHAX - YINC*NYl DO 70 IY=l,NY2 - NY1 IF<HOD<IY+NYl,101,NE,11 GOTO 50 IF<CRTPLT> GOTO 42 WRITE<LLD,1020) YT GOTO 43 42 WRITE<LLD,10401 YT 43 WRITE<LLD,10221 <WRK(IX,IY>,IX=l,NXI IF<CRTPLTI GOTO 44 WRITE<LLD,10241 YT GOTO 60 44 WRITE<LLD,1044) GOTO 60 50 IF<CRTPLTI GOTO 51 WRITE<LLD,1030) GOTO 52 51 WRITE<LLD,10501 52 WRITE<LLD,10221 <WRK(IX,IY>,IX=l,NX> WRITE<LLD,10341 60 YT= YT - YINC 70 CONTINUE IF<NY2,GE,NY> GOTO 72 NY1 = NY2 NY2 = HIN0<NY,NY2+HAXNYI GOTO 4 C PUT OUT BOTTOH XAXIS, 72 CALL XAXIS<LLD,XHIN,XINC,NX,CRTPLT,'B') 990 WRK<l,11 = ERR RETURN 00 N = ·-. IF ( M0D<IT,10),E0 ,1> J . WRITE<LLD,1012> J CONTINUE WRITE <LU1 ,1 01 4) IF (T 0PP0T,ra .· T') RETURN GOTO 10 j WRITE<LLD , 1060> XT = XMIN DO 20 I=l,NX -20,20 WRITE<LLD,1 00 2> XT XT = XT t 20,*XINC CONTINUE WRITE<LLD,1004) XT iF<TOPBOT,EO,'B'> RETURN IF<CRTPLT> GOTO 32 WRITE<LLD,101 0) GOTO 34 WRITE <LL D, 1070) DO 40 I=l,N X IT= I C C C IF<NR0W,LE,NC0LI GOTO 30 NC0U=NC0L-1 DO 20 J=1,NR0W-NC0L,NC0L L=NROW-J-NC0ll D0NE=,TRUE, DO 10 K=l,L,NC0L IF< PF LAG I GOTO 1 IF< A< Kl ,LE,A< K+ NC0LII GOTO 10 DA TA NC0L/3 / INTEGER NR0W, PFLAG REAL A(ll L0Gt.CAL*1 DONE C SORT A LINEAR ARRAY AS IF IT WHERE DIMENSIONED C NR0W*3 SO THAT THE FIRST ROW IS EITHER IN INCREASING ORDER CORIN ORDER OF ' XMIN ' ( XMAX-3*HW I C 17-JUN-78 SUPR0UTINE S0RTG(A,NR0W,PFLAG> C R, F,MURPHY C C V00 c------------------ ------------------------------------------ c-----------------------------------------------------------c S0RTG, FOR C C OUTPUT FORMATS C 1000 F0RMAT('t ' ,7X$) FORMAT< • + • ,E12,5,8X$l 1002 F0RMAT( ' t ' ,E12,5 /) 1004 F0RMAT( ' t ' ,14 X$ ) 1010 FORMAT<•+ • ,AU) 1012 F0RMAT( ' t 'I) 1014 FORMAT< ' t', 3XS> 1060 10 70 F0RMAT (' t ',1·0X$l END 40 34 32 30 20 12 14 20 30 10 5 2 GOTO 2 IF( CAOO-3, *A<K+ 2> >,L E, *CA ( KtNC0L> -3, *A< K+N C0L t2>> > GOTO 10 DO 5 I =0 ,NC0L1 Il =K+I ATEM=A<Il> A(Il>=A(IltNC0L) A< I1 +NC0L) =ATEM CONTINUE D0NE=,FALSE, CONTINUE IF<D0NE> GOTO 30 CONTINUE RETURN END L,.) (X) 84 APPENDIX B GELFIT fitting session example 85 In the example which follows, the operator first requested a cormnand summary. If more information had been needed about a particular cormnand, the user could have typed HELP/LONG to obtain it. The file GRR209.DAT was then read in for use in the standards calculation, and the LOG molecular weight calculation suppressed. (Normally, standards data are expressed as log molecula, weight versus relative mobility.) The data for a metrizamide gradient (Chapter 5, Figure 4B) were then read from a file which had been created from the output of a scintillation counter by a data collection program. The program was then requested to estimate peak positions (the GUESS command), with the constraint that each peak be greater than 1000 counts per minute. The guesses were refined by fitting, with the maximum number of iterations set to 20. As the fitting proceeded, the values of important parameters were output. A listing of the final parameter values was printed, which includes the calculated ''Mal.Weight" (in this case the peak density). A rough plot was generated on the terminal, the fit values stored in a new copy of the input file, and the program exited. a little over a minute. The entire process took 86 +DATE 2-Jul-79 • TIME 16:30:18 .RU GELFIT GELFIT V04-10B SHELP GELFIT Command Summary: The minimum . reouired letters of each command or oPtion All commands are terminated by the return key, <list> indicates a set of oPtional ar9uments, are shown, A <list> B k ADD more curves fit k BANDS wide CLOSE output file C D i,J DELETE curves i-J FIT k,1 FIT kits with 1 cits FIX i,J FIX curve i Param J GE i,x1,x2,e GENERATE data from Params GU Yfil,zfil GUESS curve Parameters H tYPe this HELP summary /L tYPe SY:GELFIT,HLP I infile 9et INPUT file /A ADD to Previous data KILL Points m-n K m,n L Ill, r, LIST Points m-n M i, j MODIFY curves i-J N 9et NEW curves 0 outfil set OUTPUT filename /LI set LIST filename /PL set PLOT filename PL <list> PLOT data and fit /C Plot in CRT format /LP Plot in LP format PO add more POINTS Q a,b set QUIT criteria S stdfil 9et STANDA~DS T i,J TYPE curves i-J U i,J UNFIX curve i Param J Wa set max relative WIDTH X a,b XSCALE <x+a>*b Y a,b YSCALE (yta>*b SSTANDARDS/NOLOG GRR209 UNP GRH202 $GUESS 1000 SFI.T 20 87 ETCSEC> 1. 2. 1, 2. 2. 2, 1. 7. 2. 3. RMS 447.9610 370.2311 321.2720 286.9051 286.1761 286.1088 286.0698 286.0692 286.0439 286,0317 MAX.REL.CHANGE IN Pt NEW VALUE CIT.I ( 6) 0.4717965E-01 0,9626734 0 ( 6) 0.1025851 0 1. 174351 ( 6) 0.2164982 1.110425 0 ( 2) 3801.431 0 0.1601599 ( 5) 1010,346 0 0.2930580E-01 0 0.9214476E-02 ( 6) 0.2216258 0 0.4176276E-02 ( 6) 0.2207003 9 0.8147353£-03 ( 6) 0.2205205 0.6298451E-03 ( 6) 0.2203816 1 0.8411340E-04 ( 6) 0,2203630 2 88 GRH202 t03 GELFIT V04-10B 02-JUL-79 PRF62.2-2 l'THY <SP.04,.112) 53 Points 6 Parameters 2 curves, Band matrix 6 Parameters wide Usins MW standards file GRR209.DAT Xmax Ymax HalfW Area Fraction Mol.Weisht 1 0.46609 3785.3 0.29404E-01 222.61 o.33224 1 .1775 2 0.59149 1015.2 0.22036 447.41 o.66776 1.1400 Area(fit) = RMS= Ymean = 670.01251 286.0317 678.30188 Area(data) = CHISQ = 2882142. ZECRMS/Ymean) Fittins terminated with Max.Rel.Chanse <= $PLOT 675.00000 = 42.169 O.lOOOE-03 Zdiff = NDF = 47 0.739 89 GRH202 103 GELFIT V04-10B 02-JUL-79 PRF62.2-2 l'THY <SP,04, -.112) O,OOOOOEtOO 0.33333EtOO 0.66667Et00 0.10000Et01 0.46Et04- :---------:---------:---------:---------:---------:---------: 0 ** ** ** ** 0.35Et04 * 0 * *O * * * 0,24Et04 * *O * * * * 0 * 0,12Et04- * 00 0 00 0,10Et03 , TIME 16l31:19 ** *** 0 0 000**0000 O*O***O 0 *O **0**00** 00** O*O* O* 00* 000 **000 ***000 0000: ****** - :---------:---------:---------:---------:---------:---------: O,OOOOOEtOO $CLOSE $EXIT 0 00 0 * * * *O * *O * * 0.33333Et00 0,66667E+OO 0,10000Et01