binaries/src/clustalo/src/squid/iupac.c

   1 /*****************************************************************
   2  * SQUID - a library of functions for biological sequence analysis
   3  * Copyright (C) 1992-2002 Washington University School of Medicine
   4  *
   5  *     This source code is freely distributed under the terms of the
   6  *     GNU General Public License. See the files COPYRIGHT and LICENSE
   7  *     for details.
   8  *****************************************************************/
   9
  10 /* iupac.c
  11  *
  12  * Globally defines the IUPAC symbols for nucleic acid sequence
  13  * Slowly evolving into a repository of globals. Tue Apr 20 1993
  14  *
  15  * RCS $Id: iupac.c 217 2011-03-19 10:27:10Z andreas $ (Original squid RCS Id: iupac.c,v 1.3 2001/02/21 21:09:10 eddy Exp)
  16  */
  17 #include "squid.h"
  18
  19 /* Default expected nucleotide occurrence frequencies, A/C/G/T.
  20  * Used (for instance) as the default distribution for
  21  * i.i.d. random nucleotide sequences.
  22  */
  23 float dnafq[4] = { 0.25, 0.25, 0.25, 0.25 };
  24
  25 /* Dayhoff f(i) amino acid occurrence frequencies.
  26  * From SwissProt 34: 21,210,388 residues
  27  * In alphabetic order by single-letter code.
  28  * Used (for instance) as the default distribution for
  29  * i.i.d. random protein sequences.
  30  */
  31 float aafq[20] = {
  32   0.075520,                     /* A */
  33   0.016973,                     /* C */
  34   0.053029,                     /* D */
  35   0.063204,                     /* E */
  36   0.040762,                     /* F */
  37   0.068448,                     /* G */
  38   0.022406,                     /* H */
  39   0.057284,                     /* I */
  40   0.059398,                     /* K */
  41   0.093399,                     /* L */
  42   0.023569,                     /* M */
  43   0.045293,                     /* N */
  44   0.049262,                     /* P */
  45   0.040231,                     /* Q */
  46   0.051573,                     /* R */
  47   0.072214,                     /* S */
  48   0.057454,                     /* T */
  49   0.065252,                     /* V */
  50   0.012513,                     /* W */
  51   0.031985                      /* Y */
  52 };
  53
  54 char aa_alphabet[] = AMINO_ALPHABET;
  55                                 /* aa_index converts to pam's 27x27 scheme */
  56 int  aa_index[20]  = { 0,  2,  3,  4,  5,  6,  7,  8, 10, 11,
  57                       12, 13, 15, 16, 17, 18, 19, 21, 22, 24 };
  58
  59                                 /* IUPAC code translations */
  60                                 /* note: sequence chars are UPPER CASE */
  61 struct iupactype iupac[] = {
  62   { 'A', 'T', NTA, NTT, },
  63   { 'C', 'G', NTC, NTG, },
  64   { 'G', 'C', NTG, NTC, },
  65   { 'T', 'A', NTT, NTA, },
  66   { 'U', 'A', NTU, NTA, },
  67   { 'N', 'N', NTN, NTN, },
  68   { ' ', ' ', NTGAP, NTGAP, },
  69   { 'R', 'Y', NTR, NTY, },
  70   { 'Y', 'R', NTY, NTR, },
  71   { 'M', 'K', NTM, NTK, },
  72   { 'K', 'M', NTK, NTM, },
  73   { 'S', 'S', NTS, NTS, },
  74   { 'W', 'W', NTW, NTW, },
  75   { 'H', 'D', NTH, NTD, },
  76   { 'B', 'V', NTB, NTV, },
  77   { 'V', 'B', NTV, NTB, },
  78   { 'D', 'H', NTD, NTH, },
  79   };
  80
  81
  82 char *stdcode1[65] = {
  83   "K",                          /* AAA */
  84   "N",                          /* AAC */
  85   "K",                          /* AAG */
  86   "N",                          /* AAU */
  87   "T",                          /* ACA */
  88   "T",                          /* ACC */
  89   "T",                          /* ACG */
  90   "T",                          /* ACU */
  91   "R",                          /* AGA */
  92   "S",                          /* AGC */
  93   "R",                          /* AGG */
  94   "S",                          /* AGU */
  95   "I",                          /* AUA */
  96   "I",                          /* AUC */
  97   "M",                          /* AUG */
  98   "I",                          /* AUU */
  99   "Q",                          /* CAA */
 100   "H",                          /* CAC */
 101   "Q",                          /* CAG */
 102   "H",                          /* CAU */
 103   "P",                          /* CCA */
 104   "P",                          /* CCC */
 105   "P",                          /* CCG */
 106   "P",                          /* CCU */
 107   "R",                          /* CGA */
 108   "R",                          /* CGC */
 109   "R",                          /* CGG */
 110   "R",                          /* CGU */
 111   "L",                          /* CUA */
 112   "L",                          /* CUC */
 113   "L",                          /* CUG */
 114   "L",                          /* CUU */
 115   "E",                          /* GAA */
 116   "D",                          /* GAC */
 117   "E",                          /* GAG */
 118   "D",                          /* GAU */
 119   "A",                          /* GCA */
 120   "A",                          /* GCC */
 121   "A",                          /* GCG */
 122   "A",                          /* GCU */
 123   "G",                          /* GGA */
 124   "G",                          /* GGC */
 125   "G",                          /* GGG */
 126   "G",                          /* GGU */
 127   "V",                          /* GUA */
 128   "V",                          /* GUC */
 129   "V",                          /* GUG */
 130   "V",                          /* GUU */
 131   "*",                          /* UAA */
 132   "Y",                          /* UAC */
 133   "*",                          /* UAG */
 134   "Y",                          /* UAU */
 135   "S",                          /* UCA */
 136   "S",                          /* UCC */
 137   "S",                          /* UCG */
 138   "S",                          /* UCU */
 139   "*",                          /* UGA */
 140   "C",                          /* UGC */
 141   "W",                          /* UGG */
 142   "C",                          /* UGU */
 143   "L",                          /* UUA */
 144   "F",                          /* UUC */
 145   "L",                          /* UUG */
 146   "F",                          /* UUU */
 147   "X",                          /* unknown */
 148 };
 149
 150
 151
 152
 153 char *stdcode3[65] = {
 154   "Lys",                        /* AAA */
 155   "Asn",                        /* AAC */
 156   "Lys",                        /* AAG */
 157   "Asn",                        /* AAU */
 158   "Thr",                        /* ACA */
 159   "Thr",                        /* ACC */
 160   "Thr",                        /* ACG */
 161   "Thr",                        /* ACU */
 162   "Arg",                        /* AGA */
 163   "Ser",                        /* AGC */
 164   "Arg",                        /* AGG */
 165   "Ser",                        /* AGU */
 166   "Ile",                        /* AUA */
 167   "Ile",                        /* AUC */
 168   "Met",                        /* AUG */
 169   "Ile",                        /* AUU */
 170   "Gln",                        /* CAA */
 171   "His",                        /* CAC */
 172   "Gln",                        /* CAG */
 173   "His",                        /* CAU */
 174   "Pro",                        /* CCA */
 175   "Pro",                        /* CCC */
 176   "Pro",                        /* CCG */
 177   "Pro",                        /* CCU */
 178   "Arg",                        /* CGA */
 179   "Arg",                        /* CGC */
 180   "Arg",                        /* CGG */
 181   "Arg",                        /* CGU */
 182   "Leu",                        /* CUA */
 183   "Leu",                        /* CUC */
 184   "Leu",                        /* CUG */
 185   "Leu",                        /* CUU */
 186   "Glu",                        /* GAA */
 187   "Asp",                        /* GAC */
 188   "Glu",                        /* GAG */
 189   "Asp",                        /* GAU */
 190   "Ala",                        /* GCA */
 191   "Ala",                        /* GCC */
 192   "Ala",                        /* GCG */
 193   "Ala",                        /* GCU */
 194   "Gly",                        /* GGA */
 195   "Gly",                        /* GGC */
 196   "Gly",                        /* GGG */
 197   "Gly",                        /* GGU */
 198   "Val",                        /* GUA */
 199   "Val",                        /* GUC */
 200   "Val",                        /* GUG */
 201   "Val",                        /* GUU */
 202   "***",                        /* UAA */
 203   "Tyr",                        /* UAC */
 204   "***",                        /* UAG */
 205   "Tyr",                        /* UAU */
 206   "Ser",                        /* UCA */
 207   "Ser",                        /* UCC */
 208   "Ser",                        /* UCG */
 209   "Ser",                        /* UCU */
 210   "***",                        /* UGA */
 211   "Cys",                        /* UGC */
 212   "Trp",                        /* UGG */
 213   "Cys",                        /* UGU */
 214   "Leu",                        /* UUA */
 215   "Phe",                        /* UUC */
 216   "Leu",                        /* UUG */
 217   "Trp",                        /* UUU */
 218   "XXX",                        /* unknown */
 219 };