#!/usr/bin/perl -w
# AA_MODULE
#    VERSION: Version 1.2 (16 Nov 2003)
#    PURPOSE:  Exports three hashes:
#          %aa_full_name     - converts 1- or 3-letter code to full name of amino acid
#          %aa_1_letter_code - converts full name or 3-letter code to 1-letter amino acid code
#          %aa_3_letter_code - converts full name or 1-letter code to 3-letter amino acid code
#       In all cases, the hash also converts an element to itself. For example:
#          $aa_full_name{"I"} ==> "Isoleucine"
#          $aa_1_letter_code{"ARG"} ==> "R"
#          $aa_3_letter_code{"PRO"} ==> "Pro"
#       The key must be all upper case

  package AA_module;
  require Exporter;
  our @ISA = qw(Exporter);
  our @EXPORT = qw(%aa_full_name  %aa_1_letter_code  %aa_3_letter_code);

############## LIBRARIES AND PRAGMAS ################

  use strict;

#################### CONSTANTS ######################

  my $true = 1;

#################### VARIABLES ######################

  our %aa_1_letter_code;       # Exported hash
  our %aa_3_letter_code;       # Exported hash
  our %aa_full_name;           # Exported hash

################# INITIALIZATION ####################

  Read_aa_info ( );
  return $true;

#################### SUBROUTINES ####################

##### READ_AA_INFO
#     Initializes hashes that interconvert different names of amino acids
#     Information is stored in a huge string in Initialize_aa_string
#     String is split up line by line, then parsed
  sub Read_aa_info {
     my $line;
     my $aa_string = Initialize_aa_string ( );   # Get all aa info

     while ($aa_string =~ /(.+?)\n/g) {          # Split up by line
        $line = $1;
        if ($line =~ /^\s*#/) {next}             # Ignore comments
        $line =~ /\s*(.+?)\W+(\w)\W+(\w\w\w)\W+(\d+)/;
        my $aa_full_name = $1;
        my $aa_1_letter_code = $2;
        my $aa_3_letter_code = $3;
        my $aa_atoms = $4;
        $aa_full_name{ uc($aa_full_name) } = $aa_full_name;
        $aa_full_name{ uc($aa_1_letter_code) } = $aa_full_name;
        $aa_full_name{ uc($aa_3_letter_code) } = $aa_full_name;
        $aa_1_letter_code{ uc($aa_full_name) } = $aa_1_letter_code;
        $aa_1_letter_code{ uc($aa_1_letter_code) } = $aa_1_letter_code;
        $aa_1_letter_code{ uc($aa_3_letter_code) } = $aa_1_letter_code;
        $aa_3_letter_code{ uc($aa_full_name) } = $aa_3_letter_code;
        $aa_3_letter_code{ uc($aa_1_letter_code) } = $aa_3_letter_code;
        $aa_3_letter_code{ uc($aa_3_letter_code) } = $aa_3_letter_code;
     }
  }

##### INITIALIZE_AA_STRING
#     Simply returns a single string with all amino acid info
  sub Initialize_aa_string {
     return "
       #    Amino acid  1-letter code  3-letter code    atoms
            alanine        A                 Ala           5
            arginine       R                 Arg          11
            asparagine     N                 Asn           8
            aspartic acid  D                 Asp           8 
            cysteine       C                 Cys           6
            glutamic acid  E                 Glu           9
            glutamine      Q                 Gln           9
            glycine        G                 Gly           4
            histidine      H                 His          10
            isoleucine     I                 Ile           8
            leucine        L                 Leu           8
            lysine         K                 Lys           9
            methionine     M                 Met           8
            phenylalanine  F                 Phe          11
            proline        P                 Pro           7
            serine         S                 Ser           6
            threonine      T                 Thr           7
            tryptophan     W                 Trp          13
            tyrosine       Y                 Tyr          11
            valine         V                 Val           7
      "
  }