Clover icon

Coverage Report

  1. Project Clover database Tue Mar 10 2026 14:58:44 GMT
  2. Package jalview.schemes

File ResidueProperties.java

 

Coverage histogram

../../img/srcFileCovDistChart0.png
0% of files have more coverage

Code metrics

74
1,833
34
1
2,610
2,350
79
0.04
53.91
34
2.32

Classes

Class Line # Actions
ResidueProperties 36 1,833 79
0.00%
 

Contributing tests

No tests hitting this source file were found.

Source view

1    /*
2    * Jalview - A Sequence Alignment Editor and Viewer ($$Version-Rel$$)
3    * Copyright (C) $$Year-Rel$$ The Jalview Authors
4    *
5    * This file is part of Jalview.
6    *
7    * Jalview is free software: you can redistribute it and/or
8    * modify it under the terms of the GNU General Public License
9    * as published by the Free Software Foundation, either version 3
10    * of the License, or (at your option) any later version.
11    *
12    * Jalview is distributed in the hope that it will be useful, but
13    * WITHOUT ANY WARRANTY; without even the implied warranty
14    * of MERCHANTABILITY or FITNESS FOR A PARTICULAR
15    * PURPOSE. See the GNU General Public License for more details.
16    *
17    * You should have received a copy of the GNU General Public License
18    * along with Jalview. If not, see <http://www.gnu.org/licenses/>.
19    * The Jalview Authors are detailed in the 'AUTHORS' file.
20    */
21    package jalview.schemes;
22   
23    import java.awt.Color;
24    import java.util.ArrayList;
25    import java.util.Arrays;
26    import java.util.Enumeration;
27    import java.util.HashMap;
28    import java.util.Hashtable;
29    import java.util.List;
30    import java.util.Locale;
31    import java.util.Map;
32    import java.util.Vector;
33   
34    import jalview.analysis.GeneticCodes;
35   
 
36    public class ResidueProperties
37    {
38    // alphabet names used in Hidden Markov Model files
39    public static final String ALPHABET_RNA = "RNA";
40   
41    public static final String ALPHABET_DNA = "DNA";
42   
43    public static final String ALPHABET_AMINO = "amino";
44   
45    // Stores residue codes/names and colours and other things
46    public static final int[] aaIndex; // aaHash version 2.1.1 and below
47   
48    public static final int[] nucleotideIndex;
49   
50    public static final int[] purinepyrimidineIndex;
51   
52    public static final int[] secondaryStructureIndex;
53   
54    public static final Map<String, Integer> aa3Hash = new HashMap<>();
55   
56    public static final Map<String, String> aa2Triplet = new HashMap<>();
57   
58    public static final Map<String, String> nucleotideName = new HashMap<>();
59   
60    public static final Map<String, String> nucleotideAmbiguityName = new HashMap<>();
61   
62    // lookup from modified amino acid (e.g. MSE) to canonical form (e.g. MET)
63    public static final Map<String, String> modifications = new HashMap<>();
64   
65    // residue background frequencies across different alphabets
66    public static final Map<String, Map<Character, Float>> backgroundFrequencies = new HashMap<>();
67   
 
68  0 toggle static
69    {
70  0 aaIndex = new int[255];
71  0 for (int i = 0; i < 255; i++)
72    {
73  0 aaIndex[i] = 23;
74    }
75   
76  0 aaIndex['A'] = 0;
77  0 aaIndex['R'] = 1;
78  0 aaIndex['N'] = 2;
79  0 aaIndex['D'] = 3;
80  0 aaIndex['C'] = 4;
81  0 aaIndex['Q'] = 5;
82  0 aaIndex['E'] = 6;
83  0 aaIndex['G'] = 7;
84  0 aaIndex['H'] = 8;
85  0 aaIndex['I'] = 9;
86  0 aaIndex['L'] = 10;
87  0 aaIndex['K'] = 11;
88  0 aaIndex['M'] = 12;
89  0 aaIndex['F'] = 13;
90  0 aaIndex['P'] = 14;
91  0 aaIndex['S'] = 15;
92  0 aaIndex['T'] = 16;
93  0 aaIndex['W'] = 17;
94  0 aaIndex['Y'] = 18;
95  0 aaIndex['V'] = 19;
96  0 aaIndex['B'] = 20;
97  0 aaIndex['Z'] = 21;
98  0 aaIndex['X'] = 22;
99  0 aaIndex['U'] = 22;
100  0 aaIndex['a'] = 0;
101  0 aaIndex['r'] = 1;
102  0 aaIndex['n'] = 2;
103  0 aaIndex['d'] = 3;
104  0 aaIndex['c'] = 4;
105  0 aaIndex['q'] = 5;
106  0 aaIndex['e'] = 6;
107  0 aaIndex['g'] = 7;
108  0 aaIndex['h'] = 8;
109  0 aaIndex['i'] = 9;
110  0 aaIndex['l'] = 10;
111  0 aaIndex['k'] = 11;
112  0 aaIndex['m'] = 12;
113  0 aaIndex['f'] = 13;
114  0 aaIndex['p'] = 14;
115  0 aaIndex['s'] = 15;
116  0 aaIndex['t'] = 16;
117  0 aaIndex['w'] = 17;
118  0 aaIndex['y'] = 18;
119  0 aaIndex['v'] = 19;
120  0 aaIndex['b'] = 20;
121  0 aaIndex['z'] = 21;
122  0 aaIndex['x'] = 22;
123  0 aaIndex['u'] = 22; // TODO: selenocystine triplet and codons needed. also
124    // extend subt. matrices
125    }
126   
127    /**
128    * maximum (gap) index for matrices involving protein alphabet
129    */
130    public final static int maxProteinIndex = 23;
131   
132    /**
133    * maximum (gap) index for matrices involving nucleotide alphabet
134    */
135    // public final static int maxNucleotideIndex = 10;
136    public final static int maxNucleotideIndex;
137   
 
138  0 toggle static
139    {
140  0 String[][] namesArray = { { "a", "Adenine" }, { "c", "Cytosine" },
141    { "g", "Guanine" },
142    { "t", "Thymine" },
143    { "u", "Uracil" },
144    { "i", "Inosine" },
145    { "x", "Xanthine" },
146    { "r", "Unknown Purine" },
147    { "y", "Unknown Pyrimidine" },
148    { "w", "Weak nucleotide (A or T)" },
149    { "s", "Strong nucleotide (G or C)" },
150    { "m", "Amino (A or C)" },
151    { "k", "Keto (G or T)" },
152    { "b", "Not A (G or C or T)" },
153    { "h", "Not G (A or C or T)" },
154    { "d", "Not C (A or G or T)" },
155    { "v", "Not T (A or G or C)" },
156    { "n", "Unknown" } };
157    // "gap" index
158  0 maxNucleotideIndex = namesArray.length;
159   
160  0 nucleotideIndex = new int[255];
161  0 for (int i = 0; i < 255; i++)
162    {
163  0 nucleotideIndex[i] = maxNucleotideIndex; // non-nucleotide symbols are all
164    // non-gap gaps.
165    }
166   
167  0 for (int i = 0; i < namesArray.length; i++)
168    {
169  0 char c = namesArray[i][0].charAt(0);
170  0 nucleotideIndex[c] = i;
171    // Character.toUpperCase is Locale insensitive
172  0 nucleotideIndex[Character.toUpperCase(c)] = i;
173  0 nucleotideName.put(namesArray[i][0], namesArray[i][1]);
174  0 nucleotideName.put(namesArray[i][0].toUpperCase(Locale.ROOT),
175    namesArray[i][1]);
176    }
177   
178    }
179   
 
180  0 toggle static
181    {
182  0 purinepyrimidineIndex = new int[255];
183  0 for (int i = 0; i < 255; i++)
184    {
185  0 purinepyrimidineIndex[i] = 3; // non-nucleotide symbols are all non-gap
186    // gaps.
187    }
188   
189  0 purinepyrimidineIndex['A'] = 0;
190  0 purinepyrimidineIndex['a'] = 0;
191  0 purinepyrimidineIndex['C'] = 1;
192  0 purinepyrimidineIndex['c'] = 1;
193  0 purinepyrimidineIndex['G'] = 0;
194  0 purinepyrimidineIndex['g'] = 0;
195  0 purinepyrimidineIndex['T'] = 1;
196  0 purinepyrimidineIndex['t'] = 1;
197  0 purinepyrimidineIndex['U'] = 1;
198  0 purinepyrimidineIndex['u'] = 1;
199  0 purinepyrimidineIndex['I'] = 2;
200  0 purinepyrimidineIndex['i'] = 2;
201  0 purinepyrimidineIndex['X'] = 2;
202  0 purinepyrimidineIndex['x'] = 2;
203  0 purinepyrimidineIndex['R'] = 0;
204  0 purinepyrimidineIndex['r'] = 0;
205  0 purinepyrimidineIndex['Y'] = 1;
206  0 purinepyrimidineIndex['y'] = 1;
207  0 purinepyrimidineIndex['N'] = 2;
208  0 purinepyrimidineIndex['n'] = 2;
209    }
210   
 
211  0 toggle static
212    {
213  0 secondaryStructureIndex = new int[255];
214  0 for (int i = 0; i < 255; i++)
215    {
216  0 secondaryStructureIndex[i] = 3;
217    }
218   
219  0 secondaryStructureIndex['H'] = 0;
220  0 secondaryStructureIndex['E'] = 1;
221  0 secondaryStructureIndex['C'] = 2;
222    }
223   
224    private static final Integer ONE = Integer.valueOf(1);
225   
226    private static final Integer ZERO = Integer.valueOf(0);
227   
 
228  0 toggle static
229    {
230  0 aa3Hash.put("ALA", ZERO);
231  0 aa3Hash.put("ARG", ONE);
232  0 aa3Hash.put("ASN", Integer.valueOf(2));
233  0 aa3Hash.put("ASP", Integer.valueOf(3)); // D
234  0 aa3Hash.put("CYS", Integer.valueOf(4));
235  0 aa3Hash.put("GLN", Integer.valueOf(5)); // Q
236  0 aa3Hash.put("GLU", Integer.valueOf(6)); // E
237  0 aa3Hash.put("GLY", Integer.valueOf(7));
238  0 aa3Hash.put("HIS", Integer.valueOf(8));
239  0 aa3Hash.put("ILE", Integer.valueOf(9));
240  0 aa3Hash.put("LEU", Integer.valueOf(10));
241  0 aa3Hash.put("LYS", Integer.valueOf(11));
242  0 aa3Hash.put("MET", Integer.valueOf(12));
243  0 aa3Hash.put("PHE", Integer.valueOf(13));
244  0 aa3Hash.put("PRO", Integer.valueOf(14));
245  0 aa3Hash.put("SER", Integer.valueOf(15));
246  0 aa3Hash.put("THR", Integer.valueOf(16));
247  0 aa3Hash.put("TRP", Integer.valueOf(17));
248  0 aa3Hash.put("TYR", Integer.valueOf(18));
249  0 aa3Hash.put("VAL", Integer.valueOf(19));
250    // IUB Nomenclature for ambiguous peptides
251  0 aa3Hash.put("ASX", Integer.valueOf(20)); // "B";
252  0 aa3Hash.put("GLX", Integer.valueOf(21)); // Z
253  0 aa3Hash.put("XAA", Integer.valueOf(22)); // X unknown
254  0 aa3Hash.put("-", Integer.valueOf(23));
255  0 aa3Hash.put("*", Integer.valueOf(23));
256  0 aa3Hash.put(".", Integer.valueOf(23));
257  0 aa3Hash.put(" ", Integer.valueOf(23));
258  0 aa3Hash.put("Gap", Integer.valueOf(23));
259  0 aa3Hash.put("UR3", Integer.valueOf(24));
260    }
261   
 
262  0 toggle static
263    {
264  0 aa2Triplet.put("A", "ALA");
265  0 aa2Triplet.put("a", "ALA");
266  0 aa2Triplet.put("R", "ARG");
267  0 aa2Triplet.put("r", "ARG");
268  0 aa2Triplet.put("N", "ASN");
269  0 aa2Triplet.put("n", "ASN");
270  0 aa2Triplet.put("D", "ASP");
271  0 aa2Triplet.put("d", "ASP");
272  0 aa2Triplet.put("C", "CYS");
273  0 aa2Triplet.put("c", "CYS");
274  0 aa2Triplet.put("Q", "GLN");
275  0 aa2Triplet.put("q", "GLN");
276  0 aa2Triplet.put("E", "GLU");
277  0 aa2Triplet.put("e", "GLU");
278  0 aa2Triplet.put("G", "GLY");
279  0 aa2Triplet.put("g", "GLY");
280  0 aa2Triplet.put("H", "HIS");
281  0 aa2Triplet.put("h", "HIS");
282  0 aa2Triplet.put("I", "ILE");
283  0 aa2Triplet.put("i", "ILE");
284  0 aa2Triplet.put("L", "LEU");
285  0 aa2Triplet.put("l", "LEU");
286  0 aa2Triplet.put("K", "LYS");
287  0 aa2Triplet.put("k", "LYS");
288  0 aa2Triplet.put("M", "MET");
289  0 aa2Triplet.put("m", "MET");
290  0 aa2Triplet.put("F", "PHE");
291  0 aa2Triplet.put("f", "PHE");
292  0 aa2Triplet.put("P", "PRO");
293  0 aa2Triplet.put("p", "PRO");
294  0 aa2Triplet.put("S", "SER");
295  0 aa2Triplet.put("s", "SER");
296  0 aa2Triplet.put("T", "THR");
297  0 aa2Triplet.put("t", "THR");
298  0 aa2Triplet.put("W", "TRP");
299  0 aa2Triplet.put("w", "TRP");
300  0 aa2Triplet.put("Y", "TYR");
301  0 aa2Triplet.put("y", "TYR");
302  0 aa2Triplet.put("V", "VAL");
303  0 aa2Triplet.put("v", "VAL");
304    }
305   
306    public static final String[] aa = { "A", "R", "N", "D", "C", "Q", "E",
307    "G", "H", "I", "L", "K", "M", "F", "P", "S", "T", "W", "Y", "V", "B",
308    "Z", "X", "_", "*", ".", " ", "U" };
309   
310    public static final Color midBlue = new Color(100, 100, 255);
311   
312    // not currently in use
313    // public static final Vector<Color> scaleColours = new Vector<Color>();
314    // static
315    // {
316    // scaleColours.addElement(new Color(114, 0, 147));
317    // scaleColours.addElement(new Color(156, 0, 98));
318    // scaleColours.addElement(new Color(190, 0, 0));
319    // scaleColours.addElement(Color.red);
320    // scaleColours.addElement(new Color(255, 125, 0));
321    // scaleColours.addElement(Color.orange);
322    // scaleColours.addElement(new Color(255, 194, 85));
323    // scaleColours.addElement(Color.yellow);
324    // scaleColours.addElement(new Color(255, 255, 181));
325    // scaleColours.addElement(Color.white);
326    // }
327   
328    public static final Color[] taylor = { new Color(204, 255, 0),
329    // A Greenish-yellowy-yellow
330    new Color(0, 0, 255), // R Blueish-bluey-blue
331    new Color(204, 0, 255), // N Blueish-reddy-blue
332    new Color(255, 0, 0), // D Reddish-reddy-red
333    new Color(255, 255, 0), // C Yellowish-yellowy-yellow
334    new Color(255, 0, 204), // Q Reddish-bluey-red
335    new Color(255, 0, 102), // E Blueish-reddy-red
336    new Color(255, 153, 0), // G Yellowy-reddy-yellow
337    new Color(0, 102, 255), // H Greenish-bluey-blue
338    new Color(102, 255, 0), // I Greenish-yellowy-green
339    new Color(51, 255, 0), // L Yellowish-greeny-green
340    new Color(102, 0, 255), // K Reddish-bluey-blue
341    new Color(0, 255, 0), // M Greenish-greeny-green
342    new Color(0, 255, 102), // F Blueish-greeny-green
343    new Color(255, 204, 0), // P Reddish-yellowy-yellow
344    new Color(255, 51, 0), // S Yellowish-reddy-red
345    new Color(255, 102, 0), // T Reddish-yellowy-red
346    new Color(0, 204, 255), // W Blueish-greeny-green
347    new Color(0, 255, 204), // Y Greenish-bluey-green
348    new Color(153, 255, 0), // V Yellowish-greeny-yellow
349    Color.white, // B
350    Color.white, // Z
351    Color.white, // X
352    Color.white, // -
353    Color.white, // *
354    Color.white // .
355    };
356   
357    public static final Color[] nucleotide = { new Color(100, 247, 63), // A
358    new Color(255, 179, 64), // C
359    new Color(235, 65, 60), // G
360    new Color(60, 136, 238), // T
361    new Color(60, 136, 238), // U
362    Color.white, // I (inosine)
363    Color.white, // X (xanthine)
364    Color.white, // R
365    Color.white, // Y
366    Color.white, // N
367    Color.white, // w
368    Color.white, // s
369    Color.white, // m
370    Color.white, // k
371    Color.white, // b
372    Color.white, // h
373    Color.white, // d
374    Color.white, // v
375    Color.white, // Gap
376    };
377   
378    // this colour scheme devised by sduce
379    public static final Color[] nucleotideAmbiguity = {
380    Color.decode("#f0fff0"), // a
381    Color.decode("#f0fff0"), // c
382    Color.decode("#f0fff0"), // g
383    Color.decode("#f0fff0"), // t
384    Color.decode("#f0fff0"), // u
385    Color.decode("#ffffff"), // i
386    Color.decode("#4f6f6f"), // x
387    Color.decode("#CD5C5C"), // r
388    Color.decode("#008000"), // y
389    Color.decode("#4682B4"), // w
390    Color.decode("#FF8C00"), // s
391    Color.decode("#9ACD32"), // m
392    Color.decode("#9932CC"), // k
393    Color.decode("#8b4513"), // b
394    Color.decode("#808080"), // h
395    Color.decode("#483D8B"), // d
396    Color.decode("#b8860b"), // v
397    Color.decode("#2f4f4f"), // n
398    Color.white, // Gap
399    };
400   
401    // Added for PurinePyrimidineColourScheme
402    public static final Color[] purinepyrimidine = { new Color(255, 131, 250), // A,
403    // G,
404    // R
405    // purines
406    // purplish/orchid
407    new Color(64, 224, 208), // C,U, T, Y pyrimidines turquoise
408    Color.white, // all other nucleotides
409    Color.white // Gap
410    };
411   
412    // Secondary structure
413    public static final Color[] secondarystructure = { Color.red, // H
414    Color.green, // E
415    Color.gray // C
416    };
417   
418    // Zappo
419    public static final Color[] zappo = { Color.pink, // A
420    midBlue, // R
421    Color.green, // N
422    Color.red, // D
423    Color.yellow, // C
424    Color.green, // Q
425    Color.red, // E
426    Color.magenta, // G
427    midBlue, // Color.red, // H
428    Color.pink, // I
429    Color.pink, // L
430    midBlue, // K
431    Color.pink, // M
432    Color.orange, // F
433    Color.magenta, // P
434    Color.green, // S
435    Color.green, // T
436    Color.orange, // W
437    Color.orange, // Y
438    Color.pink, // V
439    Color.white, // B
440    Color.white, // Z
441    Color.white, // X
442    Color.white, // -
443    Color.white, // *
444    Color.white, // .
445    Color.white // ' '
446    };
447   
448    /*
449    * flower, blossom, sunset, ocean colour schemes from geocos.
450    * See https://gecos.biotite-python.org/
451    * https://raw.githubusercontent.com/biotite-dev/biotite/master/src/biotite/sequence/graphics/color_schemes/flower.json
452    * and https://bmcbioinformatics.biomedcentral.com/articles/10.1186/s12859-020-3526-6
453    * (https://doi.org/10.1186/s12859-020-3526-6)
454    */
455    public static final Color[] flower = { new Color(177, 138, 81), // A
456    new Color(131, 191, 241), // R
457    new Color(11, 206, 198), // N
458    new Color(1, 165, 120), // D
459    new Color(255, 87, 1), // C
460    new Color(114, 149, 174), // Q
461    new Color(45, 160, 161), // E
462    new Color(177, 194, 60), // G
463    new Color(1, 148, 249), // H
464    new Color(242, 118, 99), // I
465    new Color(223, 110, 117), // L
466    new Color(127, 195, 215), // K
467    new Color(254, 157, 175), // M
468    new Color(250, 85, 157), // F
469    new Color(79, 163, 42), // P
470    new Color(180, 189, 155), // S
471    new Color(210, 181, 118), // T
472    new Color(255, 45, 237), // W
473    new Color(201, 110, 207), // Y
474    new Color(253, 153, 123), // V
475    Color.white, // B
476    Color.white, // Z
477    Color.white, // X
478    Color.white, // -
479    Color.white, // *
480    Color.white // .
481    };
482   
483    public static final Color[] blossom = { new Color(139, 196, 180), // A
484    new Color(252, 149, 2), // R
485    new Color(181, 194, 6), // N
486    new Color(95, 165, 5), // D
487    new Color(8, 147, 254), // C
488    new Color(191, 133, 39), // Q
489    new Color(219, 181, 1), // E
490    new Color(0, 211, 130), // G
491    new Color(255, 87, 1), // H
492    new Color(154, 186, 243), // I
493    new Color(205, 165, 220), // L
494    new Color(254, 165, 39), // K
495    new Color(245, 161, 184), // M
496    new Color(247, 79, 168), // F
497    new Color(16, 214, 49), // P
498    new Color(126, 157, 89), // S
499    new Color(0, 162, 156), // T
500    new Color(254, 8, 251), // W
501    new Color(255, 78, 122), // Y
502    new Color(135, 192, 228), // V
503    Color.white, // B
504    Color.white, // Z
505    Color.white, // X
506    Color.white, // -
507    Color.white, // *
508    Color.white // .
509    };
510   
511    public static final Color[] sunset = { new Color(254, 160, 253), // A
512    new Color(133, 116, 106), // R
513    new Color(171, 200, 245), // N
514    new Color(46, 123, 190), // D
515    new Color(252, 12, 254), // C
516    new Color(140, 110, 129), // Q
517    new Color(103, 120, 146), // E
518    new Color(39, 153, 255), // G
519    new Color(219, 197, 142), // H
520    new Color(250, 33, 161), // I
521    new Color(224, 30, 130), // L
522    new Color(222, 190, 204), // K
523    new Color(209, 62, 123), // M
524    new Color(255, 56, 93), // F
525    new Color(87, 102, 249), // P
526    new Color(231, 180, 253), // S
527    new Color(166, 88, 183), // T
528    new Color(255, 55, 1), // W
529    new Color(203, 83, 57), // Y
530    new Color(254, 81, 184), // V
531    Color.white, // B
532    Color.white, // Z
533    Color.white, // X
534    Color.white, // -
535    Color.white, // *
536    Color.white // .
537    };
538   
539    public static final Color[] ocean = { new Color(198, 202, 155), // A
540    new Color(12, 160, 168), // R
541    new Color(10, 223, 195), // N
542    new Color(76, 223, 161), // D
543    new Color(198, 129, 54), // C
544    new Color(139, 211, 209), // Q
545    new Color(96, 218, 201), // E
546    new Color(51, 165, 81), // G
547    new Color(0, 207, 254), // H
548    new Color(242, 186, 170), // I
549    new Color(187, 138, 131), // L
550    new Color(64, 160, 144), // K
551    new Color(164, 139, 136), // M
552    new Color(171, 136, 174), // F
553    new Color(175, 211, 101), // P
554    new Color(109, 155, 116), // S
555    new Color(141, 149, 102), // T
556    new Color(117, 138, 238), // W
557    new Color(186, 195, 252), // Y
558    new Color(233, 190, 164), // V
559    Color.white, // B
560    Color.white, // Z
561    Color.white, // X
562    Color.white, // -
563    Color.white, // *
564    Color.white // .
565    };
566   
567    // Dunno where I got these numbers from
568    public static final double[] hyd2 = { 0.62, // A
569    0.29, // R
570    -0.90, // N
571    -0.74, // D
572    1.19, // C
573    0.48, // Q
574    -0.40, // E
575    1.38, // G
576    -1.50, // H
577    1.06, // I
578    0.64, // L
579    -0.78, // K
580    0.12, // M
581    -0.85, // F
582    -2.53, // P
583    -0.18, // S
584    -0.05, // T
585    1.08, // W
586    0.81, // Y
587    0.0, // V
588    0.26, // B
589    0.0, // Z
590    0.0 // X
591    };
592   
593    public static final double[] helix = { 1.42, 0.98, 0.67, 1.01, 0.70, 1.11,
594    1.51, 0.57, 1.00, 1.08, 1.21, 1.16, 1.45, 1.13, 0.57, 0.77, 0.83,
595    1.08, 0.69, 1.06, 0.84, 1.31, 1.00, 0.0 };
596   
597    public static final double helixmin = 0.57;
598   
599    public static final double helixmax = 1.51;
600   
601    public static final double[] strand = { 0.83, 0.93, 0.89, 0.54, 1.19,
602    1.10, 0.37, 0.75, 0.87, 1.60, 1.30, 0.74, 1.05, 1.38, 0.55, 0.75,
603    1.19, 1.37, 1.47, 1.70, 0.72, 0.74, 1.0, 0.0 };
604   
605    public static final double strandmin = 0.37;
606   
607    public static final double strandmax = 1.7;
608   
609    public static final double[] turn = { 0.66, 0.95, 1.56, 1.46, 1.19, 0.98,
610    0.74, 1.56, 0.95, 0.47, 0.59, 1.01, 0.60, 0.60, 1.52, 1.43, 0.96,
611    0.96, 1.14, 0.50, 1.51, 0.86, 1.00, 0, 0 };
612   
613    public static final double turnmin = 0.47;
614   
615    public static final double turnmax = 1.56;
616   
617    public static final double[] buried = { 1.7, 0.1, 0.4, 0.4, 4.6, 0.3, 0.3,
618    1.8, 0.8, 3.1, 2.4, 0.05, 1.9, 2.2, 0.6, 0.8, 0.7, 1.6, 0.5, 2.9, 0.4,
619    0.3, 1.358, 0.00 };
620   
621    public static final double buriedmin = 0.05;
622   
623    public static final double buriedmax = 4.6;
624   
625    // This is hydropathy index
626    // Kyte, J., and Doolittle, R.F., J. Mol. Biol.
627    // 1157, 105-132, 1982
628    public static final double[] hyd = { 1.8, -4.5, -3.5, -3.5, 2.5, -3.5,
629    -3.5, -0.4, -3.2, 4.5, 3.8, -3.9, 1.9, 2.8, -1.6, -0.8, -0.7, -0.9,
630    -1.3, 4.2, -3.5, -3.5, -0.49, 0.0 };
631   
632    public static final double hydmax = 4.5;
633   
634    public static final double hydmin = -3.9;
635   
636    // public static final double hydmax = 1.38;
637    // public static final double hydmin = -2.53;
638   
639    // not currently used
640    // public static final Map<String, Color> ssHash = new Hashtable<String,
641    // Color>();
642    // static
643    // {
644    // ssHash.put("H", Color.magenta);
645    // ssHash.put("E", Color.yellow);
646    // ssHash.put("-", Color.white);
647    // ssHash.put(".", Color.white);
648    // ssHash.put("S", Color.cyan);
649    // ssHash.put("T", Color.blue);
650    // ssHash.put("G", Color.pink);
651    // ssHash.put("I", Color.pink);
652    // ssHash.put("B", Color.yellow);
653    // }
654   
655    /*
656    * new Color(60, 136, 238), // U Color.white, // I Color.white, // X
657    * Color.white, // R Color.white, // Y Color.white, // N Color.white, // Gap
658    */
659   
660    public static String STOP = "STOP";
661   
662    public static List<String> STOP_CODONS = Arrays.asList("TGA", "TAA",
663    "TAG");
664   
665    public static String START = "ATG";
666   
667    // Stores residue codes/names and colours and other things
668    public static Map<String, Map<String, Integer>> propHash = new Hashtable<>();
669   
670    public static Map<String, Integer> hydrophobic = new Hashtable<>();
671   
672    public static Map<String, Integer> polar = new Hashtable<>();
673   
674    public static Map<String, Integer> small = new Hashtable<>();
675   
676    public static Map<String, Integer> positive = new Hashtable<>();
677   
678    public static Map<String, Integer> negative = new Hashtable<>();
679   
680    public static Map<String, Integer> charged = new Hashtable<>();
681   
682    public static Map<String, Integer> aromatic = new Hashtable<>();
683   
684    public static Map<String, Integer> aliphatic = new Hashtable<>();
685   
686    public static Map<String, Integer> tiny = new Hashtable<>();
687   
688    public static Map<String, Integer> proline = new Hashtable<>();
689   
 
690  0 toggle static
691    {
692  0 hydrophobic.put("I", ONE);
693  0 hydrophobic.put("L", ONE);
694  0 hydrophobic.put("V", ONE);
695  0 hydrophobic.put("C", ONE);
696  0 hydrophobic.put("A", ONE);
697  0 hydrophobic.put("G", ONE);
698  0 hydrophobic.put("M", ONE);
699  0 hydrophobic.put("F", ONE);
700  0 hydrophobic.put("Y", ONE);
701  0 hydrophobic.put("W", ONE);
702  0 hydrophobic.put("H", ONE);
703  0 hydrophobic.put("K", ONE);
704  0 hydrophobic.put("X", ONE);
705  0 hydrophobic.put("-", ONE);
706  0 hydrophobic.put("*", ONE);
707  0 hydrophobic.put("R", ZERO);
708  0 hydrophobic.put("E", ZERO);
709  0 hydrophobic.put("Q", ZERO);
710  0 hydrophobic.put("D", ZERO);
711  0 hydrophobic.put("N", ZERO);
712  0 hydrophobic.put("S", ZERO);
713  0 hydrophobic.put("T", ONE);
714  0 hydrophobic.put("P", ZERO);
715    }
716   
 
717  0 toggle static
718    {
719  0 polar.put("Y", ONE);
720  0 polar.put("W", ONE);
721  0 polar.put("H", ONE);
722  0 polar.put("K", ONE);
723  0 polar.put("R", ONE);
724  0 polar.put("E", ONE);
725  0 polar.put("Q", ONE);
726  0 polar.put("D", ONE);
727  0 polar.put("N", ONE);
728  0 polar.put("S", ONE);
729  0 polar.put("T", ONE);
730  0 polar.put("X", ONE);
731  0 polar.put("-", ONE);
732  0 polar.put("*", ONE);
733  0 polar.put("I", ZERO);
734  0 polar.put("L", ZERO);
735  0 polar.put("V", ZERO);
736  0 polar.put("C", ZERO);
737  0 polar.put("A", ZERO);
738  0 polar.put("G", ZERO);
739  0 polar.put("M", ZERO);
740  0 polar.put("F", ZERO);
741  0 polar.put("P", ZERO);
742    }
743   
 
744  0 toggle static
745    {
746  0 small.put("I", ZERO);
747  0 small.put("L", ZERO);
748  0 small.put("V", ONE);
749  0 small.put("C", ONE);
750  0 small.put("A", ONE);
751  0 small.put("G", ONE);
752  0 small.put("M", ZERO);
753  0 small.put("F", ZERO);
754  0 small.put("Y", ZERO);
755  0 small.put("W", ZERO);
756  0 small.put("H", ZERO);
757  0 small.put("K", ZERO);
758  0 small.put("R", ZERO);
759  0 small.put("E", ZERO);
760  0 small.put("Q", ZERO);
761  0 small.put("D", ONE);
762  0 small.put("N", ONE);
763  0 small.put("S", ONE);
764  0 small.put("T", ONE);
765  0 small.put("P", ONE);
766  0 small.put("-", ONE);
767  0 small.put("*", ONE);
768    }
769   
 
770  0 toggle static
771    {
772  0 positive.put("I", ZERO);
773  0 positive.put("L", ZERO);
774  0 positive.put("V", ZERO);
775  0 positive.put("C", ZERO);
776  0 positive.put("A", ZERO);
777  0 positive.put("G", ZERO);
778  0 positive.put("M", ZERO);
779  0 positive.put("F", ZERO);
780  0 positive.put("Y", ZERO);
781  0 positive.put("W", ZERO);
782  0 positive.put("H", ONE);
783  0 positive.put("K", ONE);
784  0 positive.put("R", ONE);
785  0 positive.put("E", ZERO);
786  0 positive.put("Q", ZERO);
787  0 positive.put("D", ZERO);
788  0 positive.put("N", ZERO);
789  0 positive.put("S", ZERO);
790  0 positive.put("T", ZERO);
791  0 positive.put("P", ZERO);
792  0 positive.put("-", ONE);
793  0 positive.put("*", ONE);
794    }
795   
 
796  0 toggle static
797    {
798  0 negative.put("I", ZERO);
799  0 negative.put("L", ZERO);
800  0 negative.put("V", ZERO);
801  0 negative.put("C", ZERO);
802  0 negative.put("A", ZERO);
803  0 negative.put("G", ZERO);
804  0 negative.put("M", ZERO);
805  0 negative.put("F", ZERO);
806  0 negative.put("Y", ZERO);
807  0 negative.put("W", ZERO);
808  0 negative.put("H", ZERO);
809  0 negative.put("K", ZERO);
810  0 negative.put("R", ZERO);
811  0 negative.put("E", ONE);
812  0 negative.put("Q", ZERO);
813  0 negative.put("D", ONE);
814  0 negative.put("N", ZERO);
815  0 negative.put("S", ZERO);
816  0 negative.put("T", ZERO);
817  0 negative.put("P", ZERO);
818  0 negative.put("-", ONE);
819  0 negative.put("*", ONE);
820    }
821   
 
822  0 toggle static
823    {
824  0 charged.put("I", ZERO);
825  0 charged.put("L", ZERO);
826  0 charged.put("V", ZERO);
827  0 charged.put("C", ZERO);
828  0 charged.put("A", ZERO);
829  0 charged.put("G", ZERO);
830  0 charged.put("M", ZERO);
831  0 charged.put("F", ZERO);
832  0 charged.put("Y", ZERO);
833  0 charged.put("W", ZERO);
834  0 charged.put("H", ONE);
835  0 charged.put("K", ONE);
836  0 charged.put("R", ONE);
837  0 charged.put("E", ONE);
838  0 charged.put("Q", ZERO);
839  0 charged.put("D", ONE);
840  0 charged.put("N", ZERO); // Asparagine is polar but not
841    // charged.
842    // Alternative would be charged and
843    // negative (in basic form)?
844  0 charged.put("S", ZERO);
845  0 charged.put("T", ZERO);
846  0 charged.put("P", ZERO);
847  0 charged.put("-", ONE);
848  0 charged.put("*", ONE);
849    }
850   
 
851  0 toggle static
852    {
853  0 aromatic.put("I", ZERO);
854  0 aromatic.put("L", ZERO);
855  0 aromatic.put("V", ZERO);
856  0 aromatic.put("C", ZERO);
857  0 aromatic.put("A", ZERO);
858  0 aromatic.put("G", ZERO);
859  0 aromatic.put("M", ZERO);
860  0 aromatic.put("F", ONE);
861  0 aromatic.put("Y", ONE);
862  0 aromatic.put("W", ONE);
863  0 aromatic.put("H", ONE);
864  0 aromatic.put("K", ZERO);
865  0 aromatic.put("R", ZERO);
866  0 aromatic.put("E", ZERO);
867  0 aromatic.put("Q", ZERO);
868  0 aromatic.put("D", ZERO);
869  0 aromatic.put("N", ZERO);
870  0 aromatic.put("S", ZERO);
871  0 aromatic.put("T", ZERO);
872  0 aromatic.put("P", ZERO);
873  0 aromatic.put("-", ONE);
874  0 aromatic.put("*", ONE);
875    }
876   
 
877  0 toggle static
878    {
879  0 aliphatic.put("I", ONE);
880  0 aliphatic.put("L", ONE);
881  0 aliphatic.put("V", ONE);
882  0 aliphatic.put("C", ZERO);
883  0 aliphatic.put("A", ZERO);
884  0 aliphatic.put("G", ZERO);
885  0 aliphatic.put("M", ZERO);
886  0 aliphatic.put("F", ZERO);
887  0 aliphatic.put("Y", ZERO);
888  0 aliphatic.put("W", ZERO);
889  0 aliphatic.put("H", ZERO);
890  0 aliphatic.put("K", ZERO);
891  0 aliphatic.put("R", ZERO);
892  0 aliphatic.put("E", ZERO);
893  0 aliphatic.put("Q", ZERO);
894  0 aliphatic.put("D", ZERO);
895  0 aliphatic.put("N", ZERO);
896  0 aliphatic.put("S", ZERO);
897  0 aliphatic.put("T", ZERO);
898  0 aliphatic.put("P", ZERO);
899  0 aliphatic.put("-", ONE);
900  0 aliphatic.put("*", ONE);
901    }
902   
 
903  0 toggle static
904    {
905  0 tiny.put("I", ZERO);
906  0 tiny.put("L", ZERO);
907  0 tiny.put("V", ZERO);
908  0 tiny.put("C", ZERO);
909  0 tiny.put("A", ONE);
910  0 tiny.put("G", ONE);
911  0 tiny.put("M", ZERO);
912  0 tiny.put("F", ZERO);
913  0 tiny.put("Y", ZERO);
914  0 tiny.put("W", ZERO);
915  0 tiny.put("H", ZERO);
916  0 tiny.put("K", ZERO);
917  0 tiny.put("R", ZERO);
918  0 tiny.put("E", ZERO);
919  0 tiny.put("Q", ZERO);
920  0 tiny.put("D", ZERO);
921  0 tiny.put("N", ZERO);
922  0 tiny.put("S", ONE);
923  0 tiny.put("T", ZERO);
924  0 tiny.put("P", ZERO);
925  0 tiny.put("-", ONE);
926  0 tiny.put("*", ONE);
927    }
928   
 
929  0 toggle static
930    {
931  0 proline.put("I", ZERO);
932  0 proline.put("L", ZERO);
933  0 proline.put("V", ZERO);
934  0 proline.put("C", ZERO);
935  0 proline.put("A", ZERO);
936  0 proline.put("G", ZERO);
937  0 proline.put("M", ZERO);
938  0 proline.put("F", ZERO);
939  0 proline.put("Y", ZERO);
940  0 proline.put("W", ZERO);
941  0 proline.put("H", ZERO);
942  0 proline.put("K", ZERO);
943  0 proline.put("R", ZERO);
944  0 proline.put("E", ZERO);
945  0 proline.put("Q", ZERO);
946  0 proline.put("D", ZERO);
947  0 proline.put("N", ZERO);
948  0 proline.put("S", ZERO);
949  0 proline.put("T", ZERO);
950  0 proline.put("P", ONE);
951  0 proline.put("-", ONE);
952  0 proline.put("*", ONE);
953    }
954   
 
955  0 toggle static
956    {
957  0 propHash.put("hydrophobic", hydrophobic);
958  0 propHash.put("small", small);
959  0 propHash.put("positive", positive);
960  0 propHash.put("negative", negative);
961  0 propHash.put("charged", charged);
962  0 propHash.put("aromatic", aromatic);
963  0 propHash.put("aliphatic", aliphatic);
964  0 propHash.put("tiny", tiny);
965  0 propHash.put("proline", proline);
966  0 propHash.put("polar", polar);
967    }
 
968  0 toggle static
969    {
970  0 int[][] propMatrixF = new int[maxProteinIndex][maxProteinIndex],
971    propMatrixPos = new int[maxProteinIndex][maxProteinIndex],
972    propMatrixEpos = new int[maxProteinIndex][maxProteinIndex];
973  0 for (int i = 0; i < maxProteinIndex; i++)
974    {
975  0 int maxF = 0, maxP = 0, maxEP = 0;
976  0 String ic = "";
977  0 if (aa.length > i)
978    {
979  0 ic += aa[i];
980    }
981    else
982    {
983  0 ic = "-";
984    }
985  0 for (int j = i + 1; j < maxProteinIndex; j++)
986    {
987  0 String jc = "";
988  0 if (aa.length > j)
989    {
990  0 jc += aa[j];
991    }
992    else
993    {
994  0 jc = "-";
995    }
996  0 propMatrixF[i][j] = 0;
997  0 propMatrixPos[i][j] = 0;
998  0 propMatrixEpos[i][j] = 0;
999  0 for (String ph : propHash.keySet())
1000    {
1001  0 Map<String, Integer> pph = propHash.get(ph);
1002  0 if (pph.get(ic) != null && pph.get(jc) != null)
1003    {
1004  0 int icp = pph.get(ic).intValue(), jcp = pph.get(jc).intValue();
1005    // Still working on these definitions.
1006  0 propMatrixPos[i][j] += icp == jcp && icp > 0 ? 2 : 0;
1007  0 propMatrixPos[j][i] += icp == jcp && icp > 0 ? 2 : 0;
1008  0 propMatrixF[i][j] += icp == jcp ? 2 : 0;
1009  0 propMatrixF[j][i] += icp == jcp ? 2 : 0;
1010  0 propMatrixEpos[i][j] += icp == jcp ? (1 + icp * 2) : 0;
1011  0 propMatrixEpos[j][i] += icp == jcp ? (1 + icp * 2) : 0;
1012    }
1013    }
1014  0 if (maxF < propMatrixF[i][j])
1015    {
1016  0 maxF = propMatrixF[i][j];
1017    }
1018  0 if (maxP < propMatrixPos[i][j])
1019    {
1020  0 maxP = propMatrixPos[i][j];
1021    }
1022  0 if (maxEP < propMatrixEpos[i][j])
1023    {
1024  0 maxEP = propMatrixEpos[i][j];
1025    }
1026    }
1027  0 propMatrixF[i][i] = maxF;
1028  0 propMatrixPos[i][i] = maxP;
1029  0 propMatrixEpos[i][i] = maxEP;
1030    }
1031    }
1032   
 
1033  0 toggle private ResidueProperties()
1034    {
1035    }
1036   
 
1037  0 toggle public static double getHydmax()
1038    {
1039  0 return hydmax;
1040    }
1041   
 
1042  0 toggle public static double getHydmin()
1043    {
1044  0 return hydmin;
1045    }
1046   
 
1047  0 toggle public static double[] getHyd()
1048    {
1049  0 return hyd;
1050    }
1051   
 
1052  0 toggle public static Map<String, Integer> getAA3Hash()
1053    {
1054  0 return aa3Hash;
1055    }
1056   
 
1057  0 toggle public static String codonTranslate(String lccodon)
1058    {
1059  0 String peptide = GeneticCodes.getInstance().getStandardCodeTable()
1060    .translate(lccodon);
1061  0 if ("*".equals(peptide))
1062    {
1063  0 return "STOP";
1064    }
1065  0 return peptide;
1066    }
1067   
1068    /*
1069    * lookup of (A-Z) alternative secondary structure symbols'
1070    * equivalents in DSSP3 notation
1071    */
1072    private static char[] toDssp3State;
 
1073  0 toggle static
1074    {
1075  0 toDssp3State = new char[9]; // for 'A'-'I'; extend if needed
1076  0 Arrays.fill(toDssp3State, ' ');
1077  0 toDssp3State['B' - 'A'] = 'E';
1078  0 toDssp3State['E' - 'A'] = 'E';
1079  0 toDssp3State['G' - 'A'] = 'H';
1080  0 toDssp3State['H' - 'A'] = 'H';
1081  0 toDssp3State['I' - 'A'] = 'H';
1082    }
1083   
1084    /**
1085    * translate from other dssp secondary structure alphabets to 3-state
1086    *
1087    * @param ssString
1088    * @return ssstring
1089    */
 
1090  0 toggle public static String getDssp3state(String ssString)
1091    {
1092  0 if (ssString == null)
1093    {
1094  0 return null;
1095    }
1096  0 int lookupSize = toDssp3State.length;
1097  0 int len = ssString.length();
1098  0 char[] trans = new char[len];
1099  0 for (int i = 0; i < len; i++)
1100    {
1101  0 char c = ssString.charAt(i);
1102  0 int index = c - 'A';
1103  0 if (index < 0 || index >= lookupSize)
1104    {
1105  0 trans[i] = ' ';
1106    }
1107    else
1108    {
1109  0 trans[i] = toDssp3State[index];
1110    }
1111    }
1112  0 return new String(trans);
1113    }
1114   
 
1115  0 toggle static
1116    {
1117  0 modifications.put("MSE", "MET"); // Selenomethionine
1118    // the rest tbc; from
1119    // http://sourceforge.net/p/jmol/mailman/message/12833570/
1120    // modifications.put("CSE", "CYS"); // Selenocysteine
1121    // modifications.put("PTR", "TYR"); // Phosphotyrosine
1122    // modifications.put("SEP", "SER"); // Phosphoserine
1123    // modifications.put("HYP", "PRO"); // 4-hydroxyproline
1124    // modifications.put("5HP", "GLU"); // Pyroglutamic acid; 5-hydroxyproline
1125    // modifications.put("PCA", "GLU"); // Pyroglutamic acid
1126    // modifications.put("LYZ", "LYS"); // 5-hydroxylysine
1127   
1128    // Additional protein alphabets used in the SCOP database and PDB files
1129    // source:
1130    // https://github.com/biopython/biopython/blob/master/Bio/Data/SCOPData.py
1131  0 modifications.put("00C", "CYS");
1132  0 modifications.put("01W", "XAA");
1133  0 modifications.put("02K", "ALA");
1134  0 modifications.put("03Y", "CYS");
1135  0 modifications.put("07O", "CYS");
1136  0 modifications.put("08P", "CYS");
1137  0 modifications.put("0A0", "ASP");
1138  0 modifications.put("0A1", "TYR");
1139  0 modifications.put("0A2", "LYS");
1140  0 modifications.put("0A8", "CYS");
1141  0 modifications.put("0AA", "VAL");
1142  0 modifications.put("0AB", "VAL");
1143  0 modifications.put("0AC", "GLY");
1144  0 modifications.put("0AD", "GLY");
1145  0 modifications.put("0AF", "TRP");
1146  0 modifications.put("0AG", "LEU");
1147  0 modifications.put("0AH", "SER");
1148  0 modifications.put("0AK", "ASP");
1149  0 modifications.put("0AM", "ALA");
1150  0 modifications.put("0AP", "CYS");
1151  0 modifications.put("0AU", "UR3");
1152  0 modifications.put("0AV", "ALA");
1153  0 modifications.put("0AZ", "PRO");
1154  0 modifications.put("0BN", "PHE");
1155  0 modifications.put("0C ", "CYS");
1156  0 modifications.put("0CS", "ALA");
1157  0 modifications.put("0DC", "CYS");
1158  0 modifications.put("0DG", "GLY");
1159  0 modifications.put("0DT", "THR");
1160  0 modifications.put("0FL", "ALA");
1161  0 modifications.put("0G ", "GLY");
1162  0 modifications.put("0NC", "ALA");
1163  0 modifications.put("0SP", "ALA");
1164  0 modifications.put("0U ", "UR3");
1165  0 modifications.put("0YG", "YG");
1166  0 modifications.put("10C", "CYS");
1167  0 modifications.put("125", "UR3");
1168  0 modifications.put("126", "UR3");
1169  0 modifications.put("127", "UR3");
1170  0 modifications.put("128", "ASN");
1171  0 modifications.put("12A", "ALA");
1172  0 modifications.put("143", "CYS");
1173  0 modifications.put("175", "ASG");
1174  0 modifications.put("193", "XAA");
1175  0 modifications.put("1AP", "ALA");
1176  0 modifications.put("1MA", "ALA");
1177  0 modifications.put("1MG", "GLY");
1178  0 modifications.put("1PA", "PHE");
1179  0 modifications.put("1PI", "ALA");
1180  0 modifications.put("1PR", "ASN");
1181  0 modifications.put("1SC", "CYS");
1182  0 modifications.put("1TQ", "TRP");
1183  0 modifications.put("1TY", "TYR");
1184  0 modifications.put("1X6", "SER");
1185  0 modifications.put("200", "PHE");
1186  0 modifications.put("23F", "PHE");
1187  0 modifications.put("23S", "XAA");
1188  0 modifications.put("26B", "THR");
1189  0 modifications.put("2AD", "XAA");
1190  0 modifications.put("2AG", "ALA");
1191  0 modifications.put("2AO", "XAA");
1192  0 modifications.put("2AR", "ALA");
1193  0 modifications.put("2AS", "XAA");
1194  0 modifications.put("2AT", "THR");
1195  0 modifications.put("2AU", "UR3");
1196  0 modifications.put("2BD", "ILE");
1197  0 modifications.put("2BT", "THR");
1198  0 modifications.put("2BU", "ALA");
1199  0 modifications.put("2CO", "CYS");
1200  0 modifications.put("2DA", "ALA");
1201  0 modifications.put("2DF", "ASN");
1202  0 modifications.put("2DM", "ASN");
1203  0 modifications.put("2DO", "XAA");
1204  0 modifications.put("2DT", "THR");
1205  0 modifications.put("2EG", "GLY");
1206  0 modifications.put("2FE", "ASN");
1207  0 modifications.put("2FI", "ASN");
1208  0 modifications.put("2FM", "MET");
1209  0 modifications.put("2GT", "THR");
1210  0 modifications.put("2HF", "HIS");
1211  0 modifications.put("2LU", "LEU");
1212  0 modifications.put("2MA", "ALA");
1213  0 modifications.put("2MG", "GLY");
1214  0 modifications.put("2ML", "LEU");
1215  0 modifications.put("2MR", "ARG");
1216  0 modifications.put("2MT", "PRO");
1217  0 modifications.put("2MU", "UR3");
1218  0 modifications.put("2NT", "THR");
1219  0 modifications.put("2OM", "UR3");
1220  0 modifications.put("2OT", "THR");
1221  0 modifications.put("2PI", "XAA");
1222  0 modifications.put("2PR", "GLY");
1223  0 modifications.put("2SA", "ASN");
1224  0 modifications.put("2SI", "XAA");
1225  0 modifications.put("2ST", "THR");
1226  0 modifications.put("2TL", "THR");
1227  0 modifications.put("2TY", "TYR");
1228  0 modifications.put("2VA", "VAL");
1229  0 modifications.put("2XA", "CYS");
1230  0 modifications.put("32S", "XAA");
1231  0 modifications.put("32T", "XAA");
1232  0 modifications.put("3AH", "HIS");
1233  0 modifications.put("3AR", "XAA");
1234  0 modifications.put("3CF", "PHE");
1235  0 modifications.put("3DA", "ALA");
1236  0 modifications.put("3DR", "ASN");
1237  0 modifications.put("3GA", "ALA");
1238  0 modifications.put("3MD", "ASP");
1239  0 modifications.put("3ME", "UR3");
1240  0 modifications.put("3NF", "TYR");
1241  0 modifications.put("3QN", "LYS");
1242  0 modifications.put("3TY", "XAA");
1243  0 modifications.put("3XH", "GLY");
1244  0 modifications.put("4AC", "ASN");
1245  0 modifications.put("4BF", "TYR");
1246  0 modifications.put("4CF", "PHE");
1247  0 modifications.put("4CY", "MET");
1248  0 modifications.put("4DP", "TRP");
1249  0 modifications.put("4F3", "GYG");
1250  0 modifications.put("4FB", "PRO");
1251  0 modifications.put("4FW", "TRP");
1252  0 modifications.put("4HT", "TRP");
1253  0 modifications.put("4IN", "TRP");
1254  0 modifications.put("4MF", "ASN");
1255  0 modifications.put("4MM", "XAA");
1256  0 modifications.put("4OC", "CYS");
1257  0 modifications.put("4PC", "CYS");
1258  0 modifications.put("4PD", "CYS");
1259  0 modifications.put("4PE", "CYS");
1260  0 modifications.put("4PH", "PHE");
1261  0 modifications.put("4SC", "CYS");
1262  0 modifications.put("4SU", "UR3");
1263  0 modifications.put("4TA", "ASN");
1264  0 modifications.put("4U7", "ALA");
1265  0 modifications.put("56A", "HIS");
1266  0 modifications.put("5AA", "ALA");
1267  0 modifications.put("5AB", "ALA");
1268  0 modifications.put("5AT", "THR");
1269  0 modifications.put("5BU", "UR3");
1270  0 modifications.put("5CG", "GLY");
1271  0 modifications.put("5CM", "CYS");
1272  0 modifications.put("5CS", "CYS");
1273  0 modifications.put("5FA", "ALA");
1274  0 modifications.put("5FC", "CYS");
1275  0 modifications.put("5FU", "UR3");
1276  0 modifications.put("5HP", "GLU");
1277  0 modifications.put("5HT", "THR");
1278  0 modifications.put("5HU", "UR3");
1279  0 modifications.put("5IC", "CYS");
1280  0 modifications.put("5IT", "THR");
1281  0 modifications.put("5IU", "UR3");
1282  0 modifications.put("5MC", "CYS");
1283  0 modifications.put("5MD", "ASN");
1284  0 modifications.put("5MU", "UR3");
1285  0 modifications.put("5NC", "CYS");
1286  0 modifications.put("5PC", "CYS");
1287  0 modifications.put("5PY", "THR");
1288  0 modifications.put("5SE", "UR3");
1289  0 modifications.put("5ZA", "TWG");
1290  0 modifications.put("64T", "THR");
1291  0 modifications.put("6CL", "LYS");
1292  0 modifications.put("6CT", "THR");
1293  0 modifications.put("6CW", "TRP");
1294  0 modifications.put("6HA", "ALA");
1295  0 modifications.put("6HC", "CYS");
1296  0 modifications.put("6HG", "GLY");
1297  0 modifications.put("6HN", "LYS");
1298  0 modifications.put("6HT", "THR");
1299  0 modifications.put("6IA", "ALA");
1300  0 modifications.put("6MA", "ALA");
1301  0 modifications.put("6MC", "ALA");
1302  0 modifications.put("6MI", "ASN");
1303  0 modifications.put("6MT", "ALA");
1304  0 modifications.put("6MZ", "ASN");
1305  0 modifications.put("6OG", "GLY");
1306  0 modifications.put("70U", "UR3");
1307  0 modifications.put("7DA", "ALA");
1308  0 modifications.put("7GU", "GLY");
1309  0 modifications.put("7JA", "ILE");
1310  0 modifications.put("7MG", "GLY");
1311  0 modifications.put("8AN", "ALA");
1312  0 modifications.put("8FG", "GLY");
1313  0 modifications.put("8MG", "GLY");
1314  0 modifications.put("8OG", "GLY");
1315  0 modifications.put("9NE", "GLU");
1316  0 modifications.put("9NF", "PHE");
1317  0 modifications.put("9NR", "ARG");
1318  0 modifications.put("9NV", "VAL");
1319  0 modifications.put("A ", "ALA");
1320  0 modifications.put("A1P", "ASN");
1321  0 modifications.put("A23", "ALA");
1322  0 modifications.put("A2L", "ALA");
1323  0 modifications.put("A2M", "ALA");
1324  0 modifications.put("A34", "ALA");
1325  0 modifications.put("A35", "ALA");
1326  0 modifications.put("A38", "ALA");
1327  0 modifications.put("A39", "ALA");
1328  0 modifications.put("A3A", "ALA");
1329  0 modifications.put("A3P", "ALA");
1330  0 modifications.put("A40", "ALA");
1331  0 modifications.put("A43", "ALA");
1332  0 modifications.put("A44", "ALA");
1333  0 modifications.put("A47", "ALA");
1334  0 modifications.put("A5L", "ALA");
1335  0 modifications.put("A5M", "CYS");
1336  0 modifications.put("A5N", "ASN");
1337  0 modifications.put("A5O", "ALA");
1338  0 modifications.put("A66", "XAA");
1339  0 modifications.put("AA3", "ALA");
1340  0 modifications.put("AA4", "ALA");
1341  0 modifications.put("AAR", "ARG");
1342  0 modifications.put("AB7", "XAA");
1343  0 modifications.put("ABA", "ALA");
1344  0 modifications.put("ABR", "ALA");
1345  0 modifications.put("ABS", "ALA");
1346  0 modifications.put("ABT", "ASN");
1347  0 modifications.put("ACB", "ASP");
1348  0 modifications.put("ACL", "ARG");
1349  0 modifications.put("AD2", "ALA");
1350  0 modifications.put("ADD", "XAA");
1351  0 modifications.put("ADX", "ASN");
1352  0 modifications.put("AEA", "XAA");
1353  0 modifications.put("AEI", "ASP");
1354  0 modifications.put("AET", "ALA");
1355  0 modifications.put("AFA", "ASN");
1356  0 modifications.put("AFF", "ASN");
1357  0 modifications.put("AFG", "GLY");
1358  0 modifications.put("AGM", "ARG");
1359  0 modifications.put("AGT", "CYS");
1360  0 modifications.put("AHB", "ASN");
1361  0 modifications.put("AHH", "XAA");
1362  0 modifications.put("AHO", "ALA");
1363  0 modifications.put("AHP", "ALA");
1364  0 modifications.put("AHS", "XAA");
1365  0 modifications.put("AHT", "XAA");
1366  0 modifications.put("AIB", "ALA");
1367  0 modifications.put("AKL", "ASP");
1368  0 modifications.put("AKZ", "ASP");
1369  0 modifications.put("ALA", "ALA");
1370  0 modifications.put("ALC", "ALA");
1371  0 modifications.put("ALM", "ALA");
1372  0 modifications.put("ALN", "ALA");
1373  0 modifications.put("ALO", "THR");
1374  0 modifications.put("ALQ", "XAA");
1375  0 modifications.put("ALS", "ALA");
1376  0 modifications.put("ALT", "ALA");
1377  0 modifications.put("ALV", "ALA");
1378  0 modifications.put("ALY", "LYS");
1379  0 modifications.put("AN8", "ALA");
1380  0 modifications.put("AP7", "ALA");
1381  0 modifications.put("APE", "XAA");
1382  0 modifications.put("APH", "ALA");
1383  0 modifications.put("API", "LYS");
1384  0 modifications.put("APK", "LYS");
1385  0 modifications.put("APM", "XAA");
1386  0 modifications.put("APP", "XAA");
1387  0 modifications.put("AR2", "ARG");
1388  0 modifications.put("AR4", "GLU");
1389  0 modifications.put("AR7", "ARG");
1390  0 modifications.put("ARG", "ARG");
1391  0 modifications.put("ARM", "ARG");
1392  0 modifications.put("ARO", "ARG");
1393  0 modifications.put("ARV", "XAA");
1394  0 modifications.put("AS ", "ALA");
1395  0 modifications.put("AS2", "ASP");
1396  0 modifications.put("AS9", "XAA");
1397  0 modifications.put("ASA", "ASP");
1398  0 modifications.put("ASB", "ASP");
1399  0 modifications.put("ASI", "ASP");
1400  0 modifications.put("ASK", "ASP");
1401  0 modifications.put("ASL", "ASP");
1402  0 modifications.put("ASM", "XAA");
1403  0 modifications.put("ASN", "ASN");
1404  0 modifications.put("ASP", "ASP");
1405  0 modifications.put("ASQ", "ASP");
1406  0 modifications.put("ASU", "ASN");
1407  0 modifications.put("ASX", "ASX");
1408  0 modifications.put("ATD", "THR");
1409  0 modifications.put("ATL", "THR");
1410  0 modifications.put("ATM", "THR");
1411  0 modifications.put("AVC", "ALA");
1412  0 modifications.put("AVN", "XAA");
1413  0 modifications.put("AYA", "ALA");
1414  0 modifications.put("AYG", "AYG");
1415  0 modifications.put("AZK", "LYS");
1416  0 modifications.put("AZS", "SER");
1417  0 modifications.put("AZY", "TYR");
1418  0 modifications.put("B1F", "PHE");
1419  0 modifications.put("B1P", "ASN");
1420  0 modifications.put("B2A", "ALA");
1421  0 modifications.put("B2F", "PHE");
1422  0 modifications.put("B2I", "ILE");
1423  0 modifications.put("B2V", "VAL");
1424  0 modifications.put("B3A", "ALA");
1425  0 modifications.put("B3D", "ASP");
1426  0 modifications.put("B3E", "GLU");
1427  0 modifications.put("B3K", "LYS");
1428  0 modifications.put("B3L", "XAA");
1429  0 modifications.put("B3M", "XAA");
1430  0 modifications.put("B3Q", "XAA");
1431  0 modifications.put("B3S", "SER");
1432  0 modifications.put("B3T", "XAA");
1433  0 modifications.put("B3U", "HIS");
1434  0 modifications.put("B3X", "ASN");
1435  0 modifications.put("B3Y", "TYR");
1436  0 modifications.put("BB6", "CYS");
1437  0 modifications.put("BB7", "CYS");
1438  0 modifications.put("BB8", "PHE");
1439  0 modifications.put("BB9", "CYS");
1440  0 modifications.put("BBC", "CYS");
1441  0 modifications.put("BCS", "CYS");
1442  0 modifications.put("BE2", "XAA");
1443  0 modifications.put("BFD", "ASP");
1444  0 modifications.put("BG1", "SER");
1445  0 modifications.put("BGM", "GLY");
1446  0 modifications.put("BH2", "ASP");
1447  0 modifications.put("BHD", "ASP");
1448  0 modifications.put("BIF", "PHE");
1449  0 modifications.put("BIL", "XAA");
1450  0 modifications.put("BIU", "ILE");
1451  0 modifications.put("BJH", "XAA");
1452  0 modifications.put("BLE", "LEU");
1453  0 modifications.put("BLY", "LYS");
1454  0 modifications.put("BMP", "ASN");
1455  0 modifications.put("BMT", "THR");
1456  0 modifications.put("BNN", "PHE");
1457  0 modifications.put("BNO", "XAA");
1458  0 modifications.put("BOE", "THR");
1459  0 modifications.put("BOR", "ARG");
1460  0 modifications.put("BPE", "CYS");
1461  0 modifications.put("BRU", "UR3");
1462  0 modifications.put("BSE", "SER");
1463  0 modifications.put("BT5", "ASN");
1464  0 modifications.put("BTA", "LEU");
1465  0 modifications.put("BTC", "CYS");
1466  0 modifications.put("BTR", "TRP");
1467  0 modifications.put("BUC", "CYS");
1468  0 modifications.put("BUG", "VAL");
1469  0 modifications.put("BVP", "UR3");
1470  0 modifications.put("BZG", "ASN");
1471  0 modifications.put("C ", "CYS");
1472  0 modifications.put("C12", "TYG");
1473  0 modifications.put("C1X", "LYS");
1474  0 modifications.put("C25", "CYS");
1475  0 modifications.put("C2L", "CYS");
1476  0 modifications.put("C2S", "CYS");
1477  0 modifications.put("C31", "CYS");
1478  0 modifications.put("C32", "CYS");
1479  0 modifications.put("C34", "CYS");
1480  0 modifications.put("C36", "CYS");
1481  0 modifications.put("C37", "CYS");
1482  0 modifications.put("C38", "CYS");
1483  0 modifications.put("C3Y", "CYS");
1484  0 modifications.put("C42", "CYS");
1485  0 modifications.put("C43", "CYS");
1486  0 modifications.put("C45", "CYS");
1487  0 modifications.put("C46", "CYS");
1488  0 modifications.put("C49", "CYS");
1489  0 modifications.put("C4R", "CYS");
1490  0 modifications.put("C4S", "CYS");
1491  0 modifications.put("C5C", "CYS");
1492  0 modifications.put("C66", "XAA");
1493  0 modifications.put("C6C", "CYS");
1494  0 modifications.put("C99", "TFG");
1495  0 modifications.put("CAF", "CYS");
1496  0 modifications.put("CAL", "XAA");
1497  0 modifications.put("CAR", "CYS");
1498  0 modifications.put("CAS", "CYS");
1499  0 modifications.put("CAV", "XAA");
1500  0 modifications.put("CAY", "CYS");
1501  0 modifications.put("CB2", "CYS");
1502  0 modifications.put("CBR", "CYS");
1503  0 modifications.put("CBV", "CYS");
1504  0 modifications.put("CCC", "CYS");
1505  0 modifications.put("CCL", "LYS");
1506  0 modifications.put("CCS", "CYS");
1507  0 modifications.put("CCY", "CYG");
1508  0 modifications.put("CDE", "XAA");
1509  0 modifications.put("CDV", "XAA");
1510  0 modifications.put("CDW", "CYS");
1511  0 modifications.put("CEA", "CYS");
1512  0 modifications.put("CFL", "CYS");
1513  0 modifications.put("CFY", "FCYG"); // check
1514  0 modifications.put("CG1", "GLY");
1515  0 modifications.put("CGA", "GLU");
1516  0 modifications.put("CGU", "GLU");
1517  0 modifications.put("CH ", "CYS");
1518  0 modifications.put("CH6", "MYG");
1519  0 modifications.put("CH7", "KYG");
1520  0 modifications.put("CHF", "XAA");
1521  0 modifications.put("CHG", "XAA");
1522  0 modifications.put("CHP", "GLY");
1523  0 modifications.put("CHS", "XAA");
1524  0 modifications.put("CIR", "ARG");
1525  0 modifications.put("CJO", "GYG");
1526  0 modifications.put("CLE", "LEU");
1527  0 modifications.put("CLG", "LYS");
1528  0 modifications.put("CLH", "LYS");
1529  0 modifications.put("CLV", "AFG");
1530  0 modifications.put("CM0", "ASN");
1531  0 modifications.put("CME", "CYS");
1532  0 modifications.put("CMH", "CYS");
1533  0 modifications.put("CML", "CYS");
1534  0 modifications.put("CMR", "CYS");
1535  0 modifications.put("CMT", "CYS");
1536  0 modifications.put("CNU", "UR3");
1537  0 modifications.put("CP1", "CYS");
1538  0 modifications.put("CPC", "XAA");
1539  0 modifications.put("CPI", "XAA");
1540  0 modifications.put("CQR", "GYG");
1541  0 modifications.put("CR0", "TLG");
1542  0 modifications.put("CR2", "GYG");
1543  0 modifications.put("CR5", "GLY");
1544  0 modifications.put("CR7", "KYG");
1545  0 modifications.put("CR8", "HYG");
1546  0 modifications.put("CRF", "TWG");
1547  0 modifications.put("CRG", "THG");
1548  0 modifications.put("CRK", "MYG");
1549  0 modifications.put("CRO", "GYG");
1550  0 modifications.put("CRQ", "QYG");
1551  0 modifications.put("CRU", "EYG");
1552  0 modifications.put("CRW", "ASG");
1553  0 modifications.put("CRX", "ASG");
1554  0 modifications.put("CS0", "CYS");
1555  0 modifications.put("CS1", "CYS");
1556  0 modifications.put("CS3", "CYS");
1557  0 modifications.put("CS4", "CYS");
1558  0 modifications.put("CS8", "ASN");
1559  0 modifications.put("CSA", "CYS");
1560  0 modifications.put("CSB", "CYS");
1561  0 modifications.put("CSD", "CYS");
1562  0 modifications.put("CSE", "CYS");
1563  0 modifications.put("CSF", "CYS");
1564  0 modifications.put("CSH", "SHG");
1565  0 modifications.put("CSI", "GLY");
1566  0 modifications.put("CSJ", "CYS");
1567  0 modifications.put("CSL", "CYS");
1568  0 modifications.put("CSO", "CYS");
1569  0 modifications.put("CSP", "CYS");
1570  0 modifications.put("CSR", "CYS");
1571  0 modifications.put("CSS", "CYS");
1572  0 modifications.put("CSU", "CYS");
1573  0 modifications.put("CSW", "CYS");
1574  0 modifications.put("CSX", "CYS");
1575  0 modifications.put("CSY", "SYG");
1576  0 modifications.put("CSZ", "CYS");
1577  0 modifications.put("CTE", "TRP");
1578  0 modifications.put("CTG", "THR");
1579  0 modifications.put("CTH", "THR");
1580  0 modifications.put("CUC", "XAA");
1581  0 modifications.put("CWR", "SER");
1582  0 modifications.put("CXM", "MET");
1583  0 modifications.put("CY0", "CYS");
1584  0 modifications.put("CY1", "CYS");
1585  0 modifications.put("CY3", "CYS");
1586  0 modifications.put("CY4", "CYS");
1587  0 modifications.put("CYA", "CYS");
1588  0 modifications.put("CYD", "CYS");
1589  0 modifications.put("CYF", "CYS");
1590  0 modifications.put("CYG", "CYS");
1591  0 modifications.put("CYJ", "XAA");
1592  0 modifications.put("CYM", "CYS");
1593  0 modifications.put("CYQ", "CYS");
1594  0 modifications.put("CYR", "CYS");
1595  0 modifications.put("CYS", "CYS");
1596  0 modifications.put("CZ2", "CYS");
1597  0 modifications.put("CZO", "GYG");
1598  0 modifications.put("CZZ", "CYS");
1599  0 modifications.put("D11", "THR");
1600  0 modifications.put("D1P", "ASN");
1601  0 modifications.put("D3 ", "ASN");
1602  0 modifications.put("D33", "ASN");
1603  0 modifications.put("D3P", "GLY");
1604  0 modifications.put("D3T", "THR");
1605  0 modifications.put("D4M", "THR");
1606  0 modifications.put("D4P", "XAA");
1607  0 modifications.put("DA ", "ALA");
1608  0 modifications.put("DA2", "XAA");
1609  0 modifications.put("DAB", "ALA");
1610  0 modifications.put("DAH", "PHE");
1611  0 modifications.put("DAL", "ALA");
1612  0 modifications.put("DAR", "ARG");
1613  0 modifications.put("DAS", "ASP");
1614  0 modifications.put("DBB", "THR");
1615  0 modifications.put("DBM", "ASN");
1616  0 modifications.put("DBS", "SER");
1617  0 modifications.put("DBU", "THR");
1618  0 modifications.put("DBY", "TYR");
1619  0 modifications.put("DBZ", "ALA");
1620  0 modifications.put("DC ", "CYS");
1621  0 modifications.put("DC2", "CYS");
1622  0 modifications.put("DCG", "GLY");
1623  0 modifications.put("DCI", "XAA");
1624  0 modifications.put("DCL", "XAA");
1625  0 modifications.put("DCT", "CYS");
1626  0 modifications.put("DCY", "CYS");
1627  0 modifications.put("DDE", "HIS");
1628  0 modifications.put("DDG", "GLY");
1629  0 modifications.put("DDN", "UR3");
1630  0 modifications.put("DDX", "ASN");
1631  0 modifications.put("DFC", "CYS");
1632  0 modifications.put("DFG", "GLY");
1633  0 modifications.put("DFI", "XAA");
1634  0 modifications.put("DFO", "XAA");
1635  0 modifications.put("DFT", "ASN");
1636  0 modifications.put("DG ", "GLY");
1637  0 modifications.put("DGH", "GLY");
1638  0 modifications.put("DGI", "GLY");
1639  0 modifications.put("DGL", "GLU");
1640  0 modifications.put("DGN", "GLN");
1641  0 modifications.put("DHA", "SER");
1642  0 modifications.put("DHI", "HIS");
1643  0 modifications.put("DHL", "XAA");
1644  0 modifications.put("DHN", "VAL");
1645  0 modifications.put("DHP", "XAA");
1646  0 modifications.put("DHU", "UR3");
1647  0 modifications.put("DHV", "VAL");
1648  0 modifications.put("DI ", "ILE");
1649  0 modifications.put("DIL", "ILE");
1650  0 modifications.put("DIR", "ARG");
1651  0 modifications.put("DIV", "VAL");
1652  0 modifications.put("DLE", "LEU");
1653  0 modifications.put("DLS", "LYS");
1654  0 modifications.put("DLY", "LYS");
1655  0 modifications.put("DM0", "LYS");
1656  0 modifications.put("DMH", "ASN");
1657  0 modifications.put("DMK", "ASP");
1658  0 modifications.put("DMT", "XAA");
1659  0 modifications.put("DN ", "ASN");
1660  0 modifications.put("DNE", "LEU");
1661  0 modifications.put("DNG", "LEU");
1662  0 modifications.put("DNL", "LYS");
1663  0 modifications.put("DNM", "LEU");
1664  0 modifications.put("DNP", "ALA");
1665  0 modifications.put("DNR", "CYS");
1666  0 modifications.put("DNS", "LYS");
1667  0 modifications.put("DOA", "XAA");
1668  0 modifications.put("DOC", "CYS");
1669  0 modifications.put("DOH", "ASP");
1670  0 modifications.put("DON", "LEU");
1671  0 modifications.put("DPB", "THR");
1672  0 modifications.put("DPH", "PHE");
1673  0 modifications.put("DPL", "PRO");
1674  0 modifications.put("DPP", "ALA");
1675  0 modifications.put("DPQ", "TYR");
1676  0 modifications.put("DPR", "PRO");
1677  0 modifications.put("DPY", "ASN");
1678  0 modifications.put("DRM", "UR3");
1679  0 modifications.put("DRP", "ASN");
1680  0 modifications.put("DRT", "THR");
1681  0 modifications.put("DRZ", "ASN");
1682  0 modifications.put("DSE", "SER");
1683  0 modifications.put("DSG", "ASN");
1684  0 modifications.put("DSN", "SER");
1685  0 modifications.put("DSP", "ASP");
1686  0 modifications.put("DT ", "THR");
1687  0 modifications.put("DTH", "THR");
1688  0 modifications.put("DTR", "TRP");
1689  0 modifications.put("DTY", "TYR");
1690  0 modifications.put("DU ", "UR3");
1691  0 modifications.put("DVA", "VAL");
1692  0 modifications.put("DXD", "ASN");
1693  0 modifications.put("DXN", "ASN");
1694  0 modifications.put("DYG", "DYG");
1695  0 modifications.put("DYS", "CYS");
1696  0 modifications.put("DZM", "ALA");
1697  0 modifications.put("E ", "ALA");
1698  0 modifications.put("E1X", "ALA");
1699  0 modifications.put("ECC", "GLN");
1700  0 modifications.put("EDA", "ALA");
1701  0 modifications.put("EFC", "CYS");
1702  0 modifications.put("EHP", "PHE");
1703  0 modifications.put("EIT", "THR");
1704  0 modifications.put("ENP", "ASN");
1705  0 modifications.put("ESB", "TYR");
1706  0 modifications.put("ESC", "MET");
1707  0 modifications.put("EXB", "XAA");
1708  0 modifications.put("EXY", "LEU");
1709  0 modifications.put("EY5", "ASN");
1710  0 modifications.put("EYS", "XAA");
1711  0 modifications.put("F2F", "PHE");
1712  0 modifications.put("FA2", "ALA");
1713  0 modifications.put("FA5", "ASN");
1714  0 modifications.put("FAG", "ASN");
1715  0 modifications.put("FAI", "ASN");
1716  0 modifications.put("FB5", "ALA");
1717  0 modifications.put("FB6", "ALA");
1718  0 modifications.put("FCL", "PHE");
1719  0 modifications.put("FFD", "ASN");
1720  0 modifications.put("FGA", "GLU");
1721  0 modifications.put("FGL", "GLY");
1722  0 modifications.put("FGP", "SER");
1723  0 modifications.put("FHL", "XAA");
1724  0 modifications.put("FHO", "LYS");
1725  0 modifications.put("FHU", "UR3");
1726  0 modifications.put("FLA", "ALA");
1727  0 modifications.put("FLE", "LEU");
1728  0 modifications.put("FLT", "TYR");
1729  0 modifications.put("FME", "MET");
1730  0 modifications.put("FMG", "GLY");
1731  0 modifications.put("FMU", "ASN");
1732  0 modifications.put("FOE", "CYS");
1733  0 modifications.put("FOX", "GLY");
1734  0 modifications.put("FP9", "PRO");
1735  0 modifications.put("FPA", "PHE");
1736  0 modifications.put("FRD", "XAA");
1737  0 modifications.put("FT6", "TRP");
1738  0 modifications.put("FTR", "TRP");
1739  0 modifications.put("FTY", "TYR");
1740  0 modifications.put("FVA", "VAL");
1741  0 modifications.put("FZN", "LYS");
1742  0 modifications.put("G ", "GLY");
1743  0 modifications.put("G25", "GLY");
1744  0 modifications.put("G2L", "GLY");
1745  0 modifications.put("G2S", "GLY");
1746  0 modifications.put("G31", "GLY");
1747  0 modifications.put("G32", "GLY");
1748  0 modifications.put("G33", "GLY");
1749  0 modifications.put("G36", "GLY");
1750  0 modifications.put("G38", "GLY");
1751  0 modifications.put("G42", "GLY");
1752  0 modifications.put("G46", "GLY");
1753  0 modifications.put("G47", "GLY");
1754  0 modifications.put("G48", "GLY");
1755  0 modifications.put("G49", "GLY");
1756  0 modifications.put("G4P", "ASN");
1757  0 modifications.put("G7M", "GLY");
1758  0 modifications.put("GAO", "GLY");
1759  0 modifications.put("GAU", "GLU");
1760  0 modifications.put("GCK", "CYS");
1761  0 modifications.put("GCM", "XAA");
1762  0 modifications.put("GDP", "GLY");
1763  0 modifications.put("GDR", "GLY");
1764  0 modifications.put("GFL", "GLY");
1765  0 modifications.put("GGL", "GLU");
1766  0 modifications.put("GH3", "GLY");
1767  0 modifications.put("GHG", "GLN");
1768  0 modifications.put("GHP", "GLY");
1769  0 modifications.put("GL3", "GLY");
1770  0 modifications.put("GLH", "GLN");
1771  0 modifications.put("GLJ", "GLU");
1772  0 modifications.put("GLK", "GLU");
1773  0 modifications.put("GLM", "XAA");
1774  0 modifications.put("GLN", "GLN");
1775  0 modifications.put("GLQ", "GLU");
1776  0 modifications.put("GLU", "GLU");
1777  0 modifications.put("GLX", "GLX");
1778  0 modifications.put("GLY", "GLY");
1779  0 modifications.put("GLZ", "GLY");
1780  0 modifications.put("GMA", "GLU");
1781  0 modifications.put("GMS", "GLY");
1782  0 modifications.put("GMU", "UR3");
1783  0 modifications.put("GN7", "GLY");
1784  0 modifications.put("GND", "XAA");
1785  0 modifications.put("GNE", "ASN");
1786  0 modifications.put("GOM", "GLY");
1787  0 modifications.put("GPL", "LYS");
1788  0 modifications.put("GS ", "GLY");
1789  0 modifications.put("GSC", "GLY");
1790  0 modifications.put("GSR", "GLY");
1791  0 modifications.put("GSS", "GLY");
1792  0 modifications.put("GSU", "GLU");
1793  0 modifications.put("GT9", "CYS");
1794  0 modifications.put("GTP", "GLY");
1795  0 modifications.put("GVL", "XAA");
1796  0 modifications.put("GYC", "CYG");
1797  0 modifications.put("GYS", "SYG");
1798  0 modifications.put("H2U", "UR3");
1799  0 modifications.put("H5M", "PRO");
1800  0 modifications.put("HAC", "ALA");
1801  0 modifications.put("HAR", "ARG");
1802  0 modifications.put("HBN", "HIS");
1803  0 modifications.put("HCS", "XAA");
1804  0 modifications.put("HDP", "UR3");
1805  0 modifications.put("HEU", "UR3");
1806  0 modifications.put("HFA", "XAA");
1807  0 modifications.put("HGL", "XAA");
1808  0 modifications.put("HHI", "HIS");
1809  0 modifications.put("HHK", "AK"); // check
1810  0 modifications.put("HIA", "HIS");
1811  0 modifications.put("HIC", "HIS");
1812  0 modifications.put("HIP", "HIS");
1813  0 modifications.put("HIQ", "HIS");
1814  0 modifications.put("HIS", "HIS");
1815  0 modifications.put("HL2", "LEU");
1816  0 modifications.put("HLU", "LEU");
1817  0 modifications.put("HMR", "ARG");
1818  0 modifications.put("HOL", "ASN");
1819  0 modifications.put("HPC", "PHE");
1820  0 modifications.put("HPE", "PHE");
1821  0 modifications.put("HPH", "PHE");
1822  0 modifications.put("HPQ", "PHE");
1823  0 modifications.put("HQA", "ALA");
1824  0 modifications.put("HRG", "ARG");
1825  0 modifications.put("HRP", "TRP");
1826  0 modifications.put("HS8", "HIS");
1827  0 modifications.put("HS9", "HIS");
1828  0 modifications.put("HSE", "SER");
1829  0 modifications.put("HSL", "SER");
1830  0 modifications.put("HSO", "HIS");
1831  0 modifications.put("HTI", "CYS");
1832  0 modifications.put("HTN", "ASN");
1833  0 modifications.put("HTR", "TRP");
1834  0 modifications.put("HV5", "ALA");
1835  0 modifications.put("HVA", "VAL");
1836  0 modifications.put("HY3", "PRO");
1837  0 modifications.put("HYP", "PRO");
1838  0 modifications.put("HZP", "PRO");
1839  0 modifications.put("I ", "ILE");
1840  0 modifications.put("I2M", "ILE");
1841  0 modifications.put("I58", "LYS");
1842  0 modifications.put("I5C", "CYS");
1843  0 modifications.put("IAM", "ALA");
1844  0 modifications.put("IAR", "ARG");
1845  0 modifications.put("IAS", "ASP");
1846  0 modifications.put("IC ", "CYS");
1847  0 modifications.put("IEL", "LYS");
1848  0 modifications.put("IEY", "HYG");
1849  0 modifications.put("IG ", "GLY");
1850  0 modifications.put("IGL", "GLY");
1851  0 modifications.put("IGU", "GLY");
1852  0 modifications.put("IIC", "SHG");
1853  0 modifications.put("IIL", "ILE");
1854  0 modifications.put("ILE", "ILE");
1855  0 modifications.put("ILG", "GLU");
1856  0 modifications.put("ILX", "ILE");
1857  0 modifications.put("IMC", "CYS");
1858  0 modifications.put("IML", "ILE");
1859  0 modifications.put("IOY", "PHE");
1860  0 modifications.put("IPG", "GLY");
1861  0 modifications.put("IPN", "ASN");
1862  0 modifications.put("IRN", "ASN");
1863  0 modifications.put("IT1", "LYS");
1864  0 modifications.put("IU ", "UR3");
1865  0 modifications.put("IYR", "TYR");
1866  0 modifications.put("IYT", "THR");
1867  0 modifications.put("IZO", "MET");
1868  0 modifications.put("JJJ", "CYS");
1869  0 modifications.put("JJK", "CYS");
1870  0 modifications.put("JJL", "CYS");
1871  0 modifications.put("JW5", "ASN");
1872  0 modifications.put("K1R", "CYS");
1873  0 modifications.put("KAG", "GLY");
1874  0 modifications.put("KCX", "LYS");
1875  0 modifications.put("KGC", "LYS");
1876  0 modifications.put("KNB", "ALA");
1877  0 modifications.put("KOR", "MET");
1878  0 modifications.put("KPI", "LYS");
1879  0 modifications.put("KST", "LYS");
1880  0 modifications.put("KYQ", "LYS");
1881  0 modifications.put("L2A", "XAA");
1882  0 modifications.put("LA2", "LYS");
1883  0 modifications.put("LAA", "ASP");
1884  0 modifications.put("LAL", "ALA");
1885  0 modifications.put("LBY", "LYS");
1886  0 modifications.put("LC ", "CYS");
1887  0 modifications.put("LCA", "ALA");
1888  0 modifications.put("LCC", "ASN");
1889  0 modifications.put("LCG", "GLY");
1890  0 modifications.put("LCH", "ASN");
1891  0 modifications.put("LCK", "LYS");
1892  0 modifications.put("LCX", "LYS");
1893  0 modifications.put("LDH", "LYS");
1894  0 modifications.put("LED", "LEU");
1895  0 modifications.put("LEF", "LEU");
1896  0 modifications.put("LEH", "LEU");
1897  0 modifications.put("LEI", "VAL");
1898  0 modifications.put("LEM", "LEU");
1899  0 modifications.put("LEN", "LEU");
1900  0 modifications.put("LET", "XAA");
1901  0 modifications.put("LEU", "LEU");
1902  0 modifications.put("LEX", "LEU");
1903  0 modifications.put("LG ", "GLY");
1904  0 modifications.put("LGP", "GLY");
1905  0 modifications.put("LHC", "XAA");
1906  0 modifications.put("LHU", "UR3");
1907  0 modifications.put("LKC", "ASN");
1908  0 modifications.put("LLP", "LYS");
1909  0 modifications.put("LLY", "LYS");
1910  0 modifications.put("LME", "GLU");
1911  0 modifications.put("LMF", "LYS");
1912  0 modifications.put("LMQ", "GLN");
1913  0 modifications.put("LMS", "ASN");
1914  0 modifications.put("LP6", "LYS");
1915  0 modifications.put("LPD", "PRO");
1916  0 modifications.put("LPG", "GLY");
1917  0 modifications.put("LPL", "XAA");
1918  0 modifications.put("LPS", "SER");
1919  0 modifications.put("LSO", "XAA");
1920  0 modifications.put("LTA", "XAA");
1921  0 modifications.put("LTR", "TRP");
1922  0 modifications.put("LVG", "GLY");
1923  0 modifications.put("LVN", "VAL");
1924  0 modifications.put("LYF", "LYS");
1925  0 modifications.put("LYK", "LYS");
1926  0 modifications.put("LYM", "LYS");
1927  0 modifications.put("LYN", "LYS");
1928  0 modifications.put("LYR", "LYS");
1929  0 modifications.put("LYS", "LYS");
1930  0 modifications.put("LYX", "LYS");
1931  0 modifications.put("LYZ", "LYS");
1932  0 modifications.put("M0H", "CYS");
1933  0 modifications.put("M1G", "GLY");
1934  0 modifications.put("M2G", "GLY");
1935  0 modifications.put("M2L", "LYS");
1936  0 modifications.put("M2S", "MET");
1937  0 modifications.put("M30", "GLY");
1938  0 modifications.put("M3L", "LYS");
1939  0 modifications.put("M5M", "CYS");
1940  0 modifications.put("MA ", "ALA");
1941  0 modifications.put("MA6", "ALA");
1942  0 modifications.put("MA7", "ALA");
1943  0 modifications.put("MAA", "ALA");
1944  0 modifications.put("MAD", "ALA");
1945  0 modifications.put("MAI", "ARG");
1946  0 modifications.put("MBQ", "TYR");
1947  0 modifications.put("MBZ", "ASN");
1948  0 modifications.put("MC1", "SER");
1949  0 modifications.put("MCG", "XAA");
1950  0 modifications.put("MCL", "LYS");
1951  0 modifications.put("MCS", "CYS");
1952  0 modifications.put("MCY", "CYS");
1953  0 modifications.put("MD3", "CYS");
1954  0 modifications.put("MD6", "GLY");
1955  0 modifications.put("MDH", "XAA");
1956  0 modifications.put("MDO", "ASG");
1957  0 modifications.put("MDR", "ASN");
1958  0 modifications.put("MEA", "PHE");
1959  0 modifications.put("MED", "MET");
1960  0 modifications.put("MEG", "GLU");
1961  0 modifications.put("MEN", "ASN");
1962  0 modifications.put("MEP", "UR3");
1963  0 modifications.put("MEQ", "GLN");
1964  0 modifications.put("MET", "MET");
1965  0 modifications.put("MEU", "GLY");
1966  0 modifications.put("MF3", "XAA");
1967  0 modifications.put("MFC", "GYG");
1968  0 modifications.put("MG1", "GLY");
1969  0 modifications.put("MGG", "ARG");
1970  0 modifications.put("MGN", "GLN");
1971  0 modifications.put("MGQ", "ALA");
1972  0 modifications.put("MGV", "GLY");
1973  0 modifications.put("MGY", "GLY");
1974  0 modifications.put("MHL", "LEU");
1975  0 modifications.put("MHO", "MET");
1976  0 modifications.put("MHS", "HIS");
1977  0 modifications.put("MIA", "ALA");
1978  0 modifications.put("MIS", "SER");
1979  0 modifications.put("MK8", "LEU");
1980  0 modifications.put("ML3", "LYS");
1981  0 modifications.put("MLE", "LEU");
1982  0 modifications.put("MLL", "LEU");
1983  0 modifications.put("MLY", "LYS");
1984  0 modifications.put("MLZ", "LYS");
1985  0 modifications.put("MME", "MET");
1986  0 modifications.put("MMO", "ARG");
1987  0 modifications.put("MMT", "THR");
1988  0 modifications.put("MND", "ASN");
1989  0 modifications.put("MNL", "LEU");
1990  0 modifications.put("MNU", "UR3");
1991  0 modifications.put("MNV", "VAL");
1992  0 modifications.put("MOD", "XAA");
1993  0 modifications.put("MP8", "PRO");
1994  0 modifications.put("MPH", "XAA");
1995  0 modifications.put("MPJ", "XAA");
1996  0 modifications.put("MPQ", "GLY");
1997  0 modifications.put("MRG", "GLY");
1998  0 modifications.put("MSA", "GLY");
1999  0 modifications.put("MSE", "MET");
2000  0 modifications.put("MSL", "MET");
2001  0 modifications.put("MSO", "MET");
2002  0 modifications.put("MSP", "XAA");
2003  0 modifications.put("MT2", "MET");
2004  0 modifications.put("MTR", "THR");
2005  0 modifications.put("MTU", "ALA");
2006  0 modifications.put("MTY", "TYR");
2007  0 modifications.put("MVA", "VAL");
2008  0 modifications.put("N ", "ASN");
2009  0 modifications.put("N10", "SER");
2010  0 modifications.put("N2C", "XAA");
2011  0 modifications.put("N5I", "ASN");
2012  0 modifications.put("N5M", "CYS");
2013  0 modifications.put("N6G", "GLY");
2014  0 modifications.put("N7P", "PRO");
2015  0 modifications.put("NA8", "ALA");
2016  0 modifications.put("NAL", "ALA");
2017  0 modifications.put("NAM", "ALA");
2018  0 modifications.put("NB8", "ASN");
2019  0 modifications.put("NBQ", "TYR");
2020  0 modifications.put("NC1", "SER");
2021  0 modifications.put("NCB", "ALA");
2022  0 modifications.put("NCX", "ASN");
2023  0 modifications.put("NCY", "XAA");
2024  0 modifications.put("NDF", "PHE");
2025  0 modifications.put("NDN", "UR3");
2026  0 modifications.put("NEM", "HIS");
2027  0 modifications.put("NEP", "HIS");
2028  0 modifications.put("NF2", "ASN");
2029  0 modifications.put("NFA", "PHE");
2030  0 modifications.put("NHL", "GLU");
2031  0 modifications.put("NIT", "XAA");
2032  0 modifications.put("NIY", "TYR");
2033  0 modifications.put("NLE", "LEU");
2034  0 modifications.put("NLN", "LEU");
2035  0 modifications.put("NLO", "LEU");
2036  0 modifications.put("NLP", "LEU");
2037  0 modifications.put("NLQ", "GLN");
2038  0 modifications.put("NMC", "GLY");
2039  0 modifications.put("NMM", "ARG");
2040  0 modifications.put("NMS", "THR");
2041  0 modifications.put("NMT", "THR");
2042  0 modifications.put("NNH", "ARG");
2043  0 modifications.put("NP3", "ASN");
2044  0 modifications.put("NPH", "CYS");
2045  0 modifications.put("NPI", "ALA");
2046  0 modifications.put("NRP", "LYG");
2047  0 modifications.put("NRQ", "MYG");
2048  0 modifications.put("NSK", "XAA");
2049  0 modifications.put("NTY", "TYR");
2050  0 modifications.put("NVA", "VAL");
2051  0 modifications.put("NYC", "TWG");
2052  0 modifications.put("NYG", "NYG");
2053  0 modifications.put("NYM", "ASN");
2054  0 modifications.put("NYS", "CYS");
2055  0 modifications.put("NZH", "HIS");
2056  0 modifications.put("O12", "XAA");
2057  0 modifications.put("O2C", "ASN");
2058  0 modifications.put("O2G", "GLY");
2059  0 modifications.put("OAD", "ASN");
2060  0 modifications.put("OAS", "SER");
2061  0 modifications.put("OBF", "XAA");
2062  0 modifications.put("OBS", "XAA");
2063  0 modifications.put("OCS", "CYS");
2064  0 modifications.put("OCY", "CYS");
2065  0 modifications.put("ODP", "ASN");
2066  0 modifications.put("OHI", "HIS");
2067  0 modifications.put("OHS", "ASP");
2068  0 modifications.put("OIC", "XAA");
2069  0 modifications.put("OIP", "ILE");
2070  0 modifications.put("OLE", "XAA");
2071  0 modifications.put("OLT", "THR");
2072  0 modifications.put("OLZ", "SER");
2073  0 modifications.put("OMC", "CYS");
2074  0 modifications.put("OMG", "GLY");
2075  0 modifications.put("OMT", "MET");
2076  0 modifications.put("OMU", "UR3");
2077  0 modifications.put("ONE", "UR3");
2078  0 modifications.put("ONH", "ALA");
2079  0 modifications.put("ONL", "XAA");
2080  0 modifications.put("OPR", "ARG");
2081  0 modifications.put("ORN", "ALA");
2082  0 modifications.put("ORQ", "ARG");
2083  0 modifications.put("OSE", "SER");
2084  0 modifications.put("OTB", "XAA");
2085  0 modifications.put("OTH", "THR");
2086  0 modifications.put("OTY", "TYR");
2087  0 modifications.put("OXX", "ASP");
2088  0 modifications.put("P ", "GLY");
2089  0 modifications.put("P1L", "CYS");
2090  0 modifications.put("P1P", "ASN");
2091  0 modifications.put("P2T", "THR");
2092  0 modifications.put("P2U", "UR3");
2093  0 modifications.put("P2Y", "PRO");
2094  0 modifications.put("P5P", "ALA");
2095  0 modifications.put("PAQ", "TYR");
2096  0 modifications.put("PAS", "ASP");
2097  0 modifications.put("PAT", "TRP");
2098  0 modifications.put("PAU", "ALA");
2099  0 modifications.put("PBB", "CYS");
2100  0 modifications.put("PBF", "PHE");
2101  0 modifications.put("PBT", "ASN");
2102  0 modifications.put("PCA", "GLU");
2103  0 modifications.put("PCC", "PRO");
2104  0 modifications.put("PCE", "XAA");
2105  0 modifications.put("PCS", "PHE");
2106  0 modifications.put("PDL", "XAA");
2107  0 modifications.put("PDU", "UR3");
2108  0 modifications.put("PEC", "CYS");
2109  0 modifications.put("PF5", "PHE");
2110  0 modifications.put("PFF", "PHE");
2111  0 modifications.put("PFX", "XAA");
2112  0 modifications.put("PG1", "SER");
2113  0 modifications.put("PG7", "GLY");
2114  0 modifications.put("PG9", "GLY");
2115  0 modifications.put("PGL", "XAA");
2116  0 modifications.put("PGN", "GLY");
2117  0 modifications.put("PGP", "GLY");
2118  0 modifications.put("PGY", "GLY");
2119  0 modifications.put("PHA", "PHE");
2120  0 modifications.put("PHD", "ASP");
2121  0 modifications.put("PHE", "PHE");
2122  0 modifications.put("PHI", "PHE");
2123  0 modifications.put("PHL", "PHE");
2124  0 modifications.put("PHM", "PHE");
2125  0 modifications.put("PIA", "AYG");
2126  0 modifications.put("PIV", "XAA");
2127  0 modifications.put("PLE", "LEU");
2128  0 modifications.put("PM3", "PHE");
2129  0 modifications.put("PMT", "CYS");
2130  0 modifications.put("POM", "PRO");
2131  0 modifications.put("PPN", "PHE");
2132  0 modifications.put("PPU", "ALA");
2133  0 modifications.put("PPW", "GLY");
2134  0 modifications.put("PQ1", "ASN");
2135  0 modifications.put("PR3", "CYS");
2136  0 modifications.put("PR5", "ALA");
2137  0 modifications.put("PR9", "PRO");
2138  0 modifications.put("PRN", "ALA");
2139  0 modifications.put("PRO", "PRO");
2140  0 modifications.put("PRS", "PRO");
2141  0 modifications.put("PSA", "PHE");
2142  0 modifications.put("PSH", "HIS");
2143  0 modifications.put("PST", "THR");
2144  0 modifications.put("PSU", "UR3");
2145  0 modifications.put("PSW", "CYS");
2146  0 modifications.put("PTA", "XAA");
2147  0 modifications.put("PTH", "TYR");
2148  0 modifications.put("PTM", "TYR");
2149  0 modifications.put("PTR", "TYR");
2150  0 modifications.put("PU ", "ALA");
2151  0 modifications.put("PUY", "ASN");
2152  0 modifications.put("PVH", "HIS");
2153  0 modifications.put("PVL", "XAA");
2154  0 modifications.put("PYA", "ALA");
2155  0 modifications.put("PYO", "UR3");
2156  0 modifications.put("PYX", "CYS");
2157  0 modifications.put("PYY", "ASN");
2158  0 modifications.put("QLG", "QLG");
2159  0 modifications.put("QMM", "GLN");
2160  0 modifications.put("QPA", "CYS");
2161  0 modifications.put("QPH", "PHE");
2162  0 modifications.put("QUO", "GLY");
2163  0 modifications.put("R ", "ALA");
2164  0 modifications.put("R1A", "CYS");
2165  0 modifications.put("R4K", "TRP");
2166  0 modifications.put("RC7", "HYG");
2167  0 modifications.put("RE0", "TRP");
2168  0 modifications.put("RE3", "TRP");
2169  0 modifications.put("RIA", "ALA");
2170  0 modifications.put("RMP", "ALA");
2171  0 modifications.put("RON", "XAA");
2172  0 modifications.put("RT ", "THR");
2173  0 modifications.put("RTP", "ASN");
2174  0 modifications.put("S1H", "SER");
2175  0 modifications.put("S2C", "CYS");
2176  0 modifications.put("S2D", "ALA");
2177  0 modifications.put("S2M", "THR");
2178  0 modifications.put("S2P", "ALA");
2179  0 modifications.put("S4A", "ALA");
2180  0 modifications.put("S4C", "CYS");
2181  0 modifications.put("S4G", "GLY");
2182  0 modifications.put("S4U", "UR3");
2183  0 modifications.put("S6G", "GLY");
2184  0 modifications.put("SAC", "SER");
2185  0 modifications.put("SAH", "CYS");
2186  0 modifications.put("SAR", "GLY");
2187  0 modifications.put("SBL", "SER");
2188  0 modifications.put("SC ", "CYS");
2189  0 modifications.put("SCH", "CYS");
2190  0 modifications.put("SCS", "CYS");
2191  0 modifications.put("SCY", "CYS");
2192  0 modifications.put("SD2", "XAA");
2193  0 modifications.put("SDG", "GLY");
2194  0 modifications.put("SDP", "SER");
2195  0 modifications.put("SEB", "SER");
2196  0 modifications.put("SEC", "ALA");
2197  0 modifications.put("SEG", "ALA");
2198  0 modifications.put("SEL", "SER");
2199  0 modifications.put("SEM", "SER");
2200  0 modifications.put("SEN", "SER");
2201  0 modifications.put("SEP", "SER");
2202  0 modifications.put("SER", "SER");
2203  0 modifications.put("SET", "SER");
2204  0 modifications.put("SGB", "SER");
2205  0 modifications.put("SHC", "CYS");
2206  0 modifications.put("SHP", "GLY");
2207  0 modifications.put("SHR", "LYS");
2208  0 modifications.put("SIB", "CYS");
2209  0 modifications.put("SIC", "DC"); // check
2210  0 modifications.put("SLA", "PRO");
2211  0 modifications.put("SLR", "PRO");
2212  0 modifications.put("SLZ", "LYS");
2213  0 modifications.put("SMC", "CYS");
2214  0 modifications.put("SME", "MET");
2215  0 modifications.put("SMF", "PHE");
2216  0 modifications.put("SMP", "ALA");
2217  0 modifications.put("SMT", "THR");
2218  0 modifications.put("SNC", "CYS");
2219  0 modifications.put("SNN", "ASN");
2220  0 modifications.put("SOC", "CYS");
2221  0 modifications.put("SOS", "ASN");
2222  0 modifications.put("SOY", "SER");
2223  0 modifications.put("SPT", "THR");
2224  0 modifications.put("SRA", "ALA");
2225  0 modifications.put("SSU", "UR3");
2226  0 modifications.put("STY", "TYR");
2227  0 modifications.put("SUB", "XAA");
2228  0 modifications.put("SUI", "DG");
2229  0 modifications.put("SUN", "SER");
2230  0 modifications.put("SUR", "UR3");
2231  0 modifications.put("SVA", "SER");
2232  0 modifications.put("SVV", "SER");
2233  0 modifications.put("SVW", "SER");
2234  0 modifications.put("SVX", "SER");
2235  0 modifications.put("SVY", "SER");
2236  0 modifications.put("SVZ", "XAA");
2237  0 modifications.put("SWG", "SWG");
2238  0 modifications.put("SYS", "CYS");
2239  0 modifications.put("T ", "THR");
2240  0 modifications.put("T11", "PHE");
2241  0 modifications.put("T23", "THR");
2242  0 modifications.put("T2S", "THR");
2243  0 modifications.put("T2T", "ASN");
2244  0 modifications.put("T31", "UR3");
2245  0 modifications.put("T32", "THR");
2246  0 modifications.put("T36", "THR");
2247  0 modifications.put("T37", "THR");
2248  0 modifications.put("T38", "THR");
2249  0 modifications.put("T39", "THR");
2250  0 modifications.put("T3P", "THR");
2251  0 modifications.put("T41", "THR");
2252  0 modifications.put("T48", "THR");
2253  0 modifications.put("T49", "THR");
2254  0 modifications.put("T4S", "THR");
2255  0 modifications.put("T5O", "UR3");
2256  0 modifications.put("T5S", "THR");
2257  0 modifications.put("T66", "XAA");
2258  0 modifications.put("T6A", "ALA");
2259  0 modifications.put("TA3", "THR");
2260  0 modifications.put("TA4", "XAA");
2261  0 modifications.put("TAF", "THR");
2262  0 modifications.put("TAL", "ASN");
2263  0 modifications.put("TAV", "ASP");
2264  0 modifications.put("TBG", "VAL");
2265  0 modifications.put("TBM", "THR");
2266  0 modifications.put("TC1", "CYS");
2267  0 modifications.put("TCP", "THR");
2268  0 modifications.put("TCQ", "TYR");
2269  0 modifications.put("TCR", "TRP");
2270  0 modifications.put("TCY", "ALA");
2271  0 modifications.put("TDD", "LEU");
2272  0 modifications.put("TDY", "THR");
2273  0 modifications.put("TFE", "THR");
2274  0 modifications.put("TFO", "ALA");
2275  0 modifications.put("TFQ", "PHE");
2276  0 modifications.put("TFT", "THR");
2277  0 modifications.put("TGP", "GLY");
2278  0 modifications.put("TH6", "THR");
2279  0 modifications.put("THC", "THR");
2280  0 modifications.put("THO", "XAA");
2281  0 modifications.put("THR", "THR");
2282  0 modifications.put("THX", "ASN");
2283  0 modifications.put("THZ", "ARG");
2284  0 modifications.put("TIH", "ALA");
2285  0 modifications.put("TLB", "ASN");
2286  0 modifications.put("TLC", "THR");
2287  0 modifications.put("TLN", "UR3");
2288  0 modifications.put("TMB", "THR");
2289  0 modifications.put("TMD", "THR");
2290  0 modifications.put("TNB", "CYS");
2291  0 modifications.put("TNR", "SER");
2292  0 modifications.put("TOX", "TRP");
2293  0 modifications.put("TP1", "THR");
2294  0 modifications.put("TPC", "CYS");
2295  0 modifications.put("TPG", "GLY");
2296  0 modifications.put("TPH", "XAA");
2297  0 modifications.put("TPL", "TRP");
2298  0 modifications.put("TPO", "THR");
2299  0 modifications.put("TPQ", "TYR");
2300  0 modifications.put("TQI", "TRP");
2301  0 modifications.put("TQQ", "TRP");
2302  0 modifications.put("TRF", "TRP");
2303  0 modifications.put("TRG", "LYS");
2304  0 modifications.put("TRN", "TRP");
2305  0 modifications.put("TRO", "TRP");
2306  0 modifications.put("TRP", "TRP");
2307  0 modifications.put("TRQ", "TRP");
2308  0 modifications.put("TRW", "TRP");
2309  0 modifications.put("TRX", "TRP");
2310  0 modifications.put("TS ", "ASN");
2311  0 modifications.put("TST", "XAA");
2312  0 modifications.put("TT ", "ASN");
2313  0 modifications.put("TTD", "THR");
2314  0 modifications.put("TTI", "UR3");
2315  0 modifications.put("TTM", "THR");
2316  0 modifications.put("TTQ", "TRP");
2317  0 modifications.put("TTS", "TYR");
2318  0 modifications.put("TY1", "TYR");
2319  0 modifications.put("TY2", "TYR");
2320  0 modifications.put("TY3", "TYR");
2321  0 modifications.put("TY5", "TYR");
2322  0 modifications.put("TYB", "TYR");
2323  0 modifications.put("TYI", "TYR");
2324  0 modifications.put("TYJ", "TYR");
2325  0 modifications.put("TYN", "TYR");
2326  0 modifications.put("TYO", "TYR");
2327  0 modifications.put("TYQ", "TYR");
2328  0 modifications.put("TYR", "TYR");
2329  0 modifications.put("TYS", "TYR");
2330  0 modifications.put("TYT", "TYR");
2331  0 modifications.put("TYU", "ASN");
2332  0 modifications.put("TYW", "TYR");
2333  0 modifications.put("TYX", "XAA");
2334  0 modifications.put("TYY", "TYR");
2335  0 modifications.put("TZB", "XAA");
2336  0 modifications.put("TZO", "XAA");
2337  0 modifications.put("U ", "UR3");
2338  0 modifications.put("U25", "UR3");
2339  0 modifications.put("U2L", "UR3");
2340  0 modifications.put("U2N", "UR3");
2341  0 modifications.put("U2P", "UR3");
2342  0 modifications.put("U31", "UR3");
2343  0 modifications.put("U33", "UR3");
2344  0 modifications.put("U34", "UR3");
2345  0 modifications.put("U36", "UR3");
2346  0 modifications.put("U37", "UR3");
2347  0 modifications.put("U8U", "UR3");
2348  0 modifications.put("UAR", "UR3");
2349  0 modifications.put("UCL", "UR3");
2350  0 modifications.put("UD5", "UR3");
2351  0 modifications.put("UDP", "ASN");
2352  0 modifications.put("UFP", "ASN");
2353  0 modifications.put("UFR", "UR3");
2354  0 modifications.put("UFT", "UR3");
2355  0 modifications.put("UMA", "ALA");
2356  0 modifications.put("UMP", "UR3");
2357  0 modifications.put("UMS", "UR3");
2358  0 modifications.put("UN1", "XAA");
2359  0 modifications.put("UN2", "XAA");
2360  0 modifications.put("UNK", "XAA");
2361  0 modifications.put("UR3", "UR3");
2362  0 modifications.put("URD", "UR3");
2363  0 modifications.put("US1", "UR3");
2364  0 modifications.put("US2", "UR3");
2365  0 modifications.put("US3", "THR");
2366  0 modifications.put("US5", "UR3");
2367  0 modifications.put("USM", "UR3");
2368  0 modifications.put("VAD", "VAL");
2369  0 modifications.put("VAF", "VAL");
2370  0 modifications.put("VAL", "VAL");
2371  0 modifications.put("VB1", "LYS");
2372  0 modifications.put("VDL", "XAA");
2373  0 modifications.put("VLL", "XAA");
2374  0 modifications.put("VLM", "XAA");
2375  0 modifications.put("VMS", "XAA");
2376  0 modifications.put("VOL", "XAA");
2377  0 modifications.put("WCR", "GYG");
2378  0 modifications.put("X ", "GLY");
2379  0 modifications.put("X2W", "GLU");
2380  0 modifications.put("X4A", "ASN");
2381  0 modifications.put("X9Q", "AFG");
2382  0 modifications.put("XAD", "ALA");
2383  0 modifications.put("XAE", "ASN");
2384  0 modifications.put("XAL", "ALA");
2385  0 modifications.put("XAR", "ASN");
2386  0 modifications.put("XCL", "CYS");
2387  0 modifications.put("XCN", "CYS");
2388  0 modifications.put("XCP", "XAA");
2389  0 modifications.put("XCR", "CYS");
2390  0 modifications.put("XCS", "ASN");
2391  0 modifications.put("XCT", "CYS");
2392  0 modifications.put("XCY", "CYS");
2393  0 modifications.put("XGA", "ASN");
2394  0 modifications.put("XGL", "GLY");
2395  0 modifications.put("XGR", "GLY");
2396  0 modifications.put("XGU", "GLY");
2397  0 modifications.put("XPR", "PRO");
2398  0 modifications.put("XSN", "ASN");
2399  0 modifications.put("XTH", "THR");
2400  0 modifications.put("XTL", "THR");
2401  0 modifications.put("XTR", "THR");
2402  0 modifications.put("XTS", "GLY");
2403  0 modifications.put("XTY", "ASN");
2404  0 modifications.put("XUA", "ALA");
2405  0 modifications.put("XUG", "GLY");
2406  0 modifications.put("XX1", "LYS");
2407  0 modifications.put("XXY", "THG");
2408  0 modifications.put("XYG", "DYG");
2409  0 modifications.put("Y ", "ALA");
2410  0 modifications.put("YCM", "CYS");
2411  0 modifications.put("YG ", "GLY");
2412  0 modifications.put("YOF", "TYR");
2413  0 modifications.put("YRR", "ASN");
2414  0 modifications.put("YYG", "GLY");
2415  0 modifications.put("Z ", "CYS");
2416  0 modifications.put("Z01", "ALA");
2417  0 modifications.put("ZAD", "ALA");
2418  0 modifications.put("ZAL", "ALA");
2419  0 modifications.put("ZBC", "CYS");
2420  0 modifications.put("ZBU", "UR3");
2421  0 modifications.put("ZCL", "PHE");
2422  0 modifications.put("ZCY", "CYS");
2423  0 modifications.put("ZDU", "UR3");
2424  0 modifications.put("ZFB", "XAA");
2425  0 modifications.put("ZGU", "GLY");
2426  0 modifications.put("ZHP", "ASN");
2427  0 modifications.put("ZTH", "THR");
2428  0 modifications.put("ZU0", "THR");
2429  0 modifications.put("ZZJ", "ALA");
2430   
2431    }
2432   
 
2433  0 toggle static
2434    {
2435  0 Map<Character, Float> amino = new HashMap<>();
2436  0 amino.put('A', 0.0826f);
2437  0 amino.put('Q', 0.0393f);
2438  0 amino.put('L', 0.0965f);
2439  0 amino.put('S', 0.0661f);
2440  0 amino.put('R', 0.0553f);
2441  0 amino.put('E', 0.0674f);
2442  0 amino.put('K', 0.0582f);
2443  0 amino.put('T', 0.0535f);
2444  0 amino.put('N', 0.0406f);
2445  0 amino.put('G', 0.0708f);
2446  0 amino.put('M', 0.0241f);
2447  0 amino.put('W', 0.0109f);
2448  0 amino.put('D', 0.0546f);
2449  0 amino.put('H', 0.0227f);
2450  0 amino.put('F', 0.0386f);
2451  0 amino.put('Y', 0.0292f);
2452  0 amino.put('C', 0.0137f);
2453  0 amino.put('I', 0.0593f);
2454  0 amino.put('P', 0.0472f);
2455  0 amino.put('V', 0.0686f);
2456  0 backgroundFrequencies.put(ALPHABET_AMINO, amino);
2457    // todo: these don't match https://www.ebi.ac.uk/uniprot/TrEMBLstats - what
2458    // are they?
2459    }
2460   
2461    // TODO get correct frequencies
2462   
 
2463  0 toggle static
2464    {
2465  0 Map<Character, Float> dna = new HashMap<>();
2466  0 dna.put('A', 0.25f);
2467  0 dna.put('C', 0.25f);
2468  0 dna.put('T', 0.25f);
2469  0 dna.put('G', 0.25f);
2470  0 backgroundFrequencies.put(ALPHABET_DNA, dna);
2471   
2472    }
2473   
 
2474  0 toggle static
2475    {
2476  0 Map<Character, Float> rna = new HashMap<>();
2477  0 rna.put('A', 0.25f);
2478  0 rna.put('C', 0.25f);
2479  0 rna.put('T', 0.25f);
2480  0 rna.put('G', 0.25f);
2481  0 backgroundFrequencies.put(ALPHABET_RNA, rna);
2482   
2483    }
2484   
 
2485  0 toggle public static String getCanonicalAminoAcid(String aA)
2486    {
2487  0 String canonical = modifications.get(aA);
2488  0 return canonical == null ? aA : canonical;
2489    }
2490   
2491    // main method generates perl representation of residue property hash
2492    // / cut here
2493    /**
2494    * @j2sIgnore
2495    * @param args
2496    */
 
2497  0 toggle public static void main(String[] args)
2498    {
2499  0 Hashtable<String, Vector<String>> aaProps = new Hashtable<>();
2500  0 jalview.bin.Console.outPrintln("my %aa = {");
2501    // invert property hashes
2502  0 for (String pname : propHash.keySet())
2503    {
2504  0 Map<String, Integer> phash = propHash.get(pname);
2505  0 for (String rname : phash.keySet())
2506    {
2507  0 Vector<String> aprops = aaProps.get(rname);
2508  0 if (aprops == null)
2509    {
2510  0 aprops = new Vector<>();
2511  0 aaProps.put(rname, aprops);
2512    }
2513  0 Integer hasprop = phash.get(rname);
2514  0 if (hasprop.intValue() == 1)
2515    {
2516  0 aprops.addElement(pname);
2517    }
2518    }
2519    }
2520  0 Enumeration<String> res = aaProps.keys();
2521  0 while (res.hasMoreElements())
2522    {
2523  0 String rname = res.nextElement();
2524   
2525  0 System.out.print("'" + rname + "' => [");
2526  0 Enumeration<String> props = aaProps.get(rname).elements();
2527  0 while (props.hasMoreElements())
2528    {
2529  0 System.out.print("'" + props.nextElement() + "'");
2530  0 if (props.hasMoreElements())
2531    {
2532  0 jalview.bin.Console.outPrintln(", ");
2533    }
2534    }
2535  0 jalview.bin.Console
2536  0 .outPrintln("]" + (res.hasMoreElements() ? "," : ""));
2537    }
2538  0 jalview.bin.Console.outPrintln("};");
2539    }
2540   
2541    // to here
2542   
2543    /**
2544    * Returns a list of residue characters for the specified inputs
2545    *
2546    * @param forNucleotide
2547    * @param includeAmbiguous
2548    * @return
2549    */
 
2550  0 toggle public static List<String> getResidues(boolean forNucleotide,
2551    boolean includeAmbiguous)
2552    {
2553  0 List<String> result = new ArrayList<>();
2554  0 if (forNucleotide)
2555    {
2556  0 for (String nuc : nucleotideName.keySet())
2557    {
2558  0 int val = nucleotideIndex[nuc.charAt(0)];
2559  0 if ((!includeAmbiguous && val > 4) || (val >= maxNucleotideIndex))
2560    {
2561  0 continue;
2562    }
2563  0 nuc = nuc.toUpperCase(Locale.ROOT);
2564  0 if (!result.contains(nuc))
2565    {
2566  0 result.add(nuc);
2567    }
2568    }
2569    }
2570    else
2571    {
2572    /*
2573    * Peptide
2574    */
2575  0 for (String res : aa3Hash.keySet())
2576    {
2577  0 int index = aa3Hash.get(res).intValue();
2578  0 if ((!includeAmbiguous && index >= 20) || index >= maxProteinIndex)
2579    {
2580  0 continue;
2581    }
2582  0 res = res.toUpperCase(Locale.ROOT);
2583  0 if (!result.contains(res))
2584    {
2585  0 result.add(res);
2586    }
2587    }
2588    }
2589   
2590  0 return result;
2591    }
2592   
2593    /**
2594    * Returns the single letter code for a three letter code, or '0' if not known
2595    *
2596    * @param threeLetterCode
2597    * not case sensitive
2598    * @return
2599    */
 
2600  0 toggle public static char getSingleCharacterCode(String threeLetterCode)
2601    {
2602  0 if (threeLetterCode == null)
2603    {
2604  0 return '0';
2605    }
2606  0 Integer index = ResidueProperties.aa3Hash
2607    .get(threeLetterCode.toUpperCase(Locale.ROOT));
2608  0 return index == null ? '0' : aa[index].charAt(0);
2609    }
2610    }