| 1 | 1 <html> | 
|  | 2 <head> | 
|  | 3 <title>MayaChemTools:Code:AminoAcids.pm</title> | 
|  | 4 <meta http-equiv="content-type" content="text/html;charset=utf-8"> | 
|  | 5 <link rel="stylesheet" type="text/css" href="../../../css/MayaChemToolsCode.css"> | 
|  | 6 </head> | 
|  | 7 <body leftmargin="20" rightmargin="20" topmargin="10" bottommargin="10"> | 
|  | 8 <br/> | 
|  | 9 <center> | 
|  | 10 <a href="http://www.mayachemtools.org" title="MayaChemTools Home"><img src="../../../images/MayaChemToolsLogo.gif" border="0" alt="MayaChemTools"></a> | 
|  | 11 </center> | 
|  | 12 <br/> | 
|  | 13 <pre> | 
|  | 14 <a name="package-AminoAcids-"></a>   1 <span class="k">package </span><span class="i">AminoAcids</span><span class="sc">;</span> | 
|  | 15    2 <span class="c">#</span> | 
|  | 16    3 <span class="c"># $RCSfile: AminoAcids.pm,v $</span> | 
|  | 17    4 <span class="c"># $Date: 2015/02/28 20:47:02 $</span> | 
|  | 18    5 <span class="c"># $Revision: 1.25 $</span> | 
|  | 19    6 <span class="c">#</span> | 
|  | 20    7 <span class="c"># Author: Manish Sud <msud@san.rr.com></span> | 
|  | 21    8 <span class="c">#</span> | 
|  | 22    9 <span class="c"># Copyright (C) 2015 Manish Sud. All rights reserved.</span> | 
|  | 23   10 <span class="c">#</span> | 
|  | 24   11 <span class="c"># This file is part of MayaChemTools.</span> | 
|  | 25   12 <span class="c">#</span> | 
|  | 26   13 <span class="c"># MayaChemTools is free software; you can redistribute it and/or modify it under</span> | 
|  | 27   14 <span class="c"># the terms of the GNU Lesser General Public License as published by the Free</span> | 
|  | 28   15 <span class="c"># Software Foundation; either version 3 of the License, or (at your option) any</span> | 
|  | 29   16 <span class="c"># later version.</span> | 
|  | 30   17 <span class="c">#</span> | 
|  | 31   18 <span class="c"># MayaChemTools is distributed in the hope that it will be useful, but without</span> | 
|  | 32   19 <span class="c"># any warranty; without even the implied warranty of merchantability of fitness</span> | 
|  | 33   20 <span class="c"># for a particular purpose.  See the GNU Lesser General Public License for more</span> | 
|  | 34   21 <span class="c"># details.</span> | 
|  | 35   22 <span class="c">#</span> | 
|  | 36   23 <span class="c"># You should have received a copy of the GNU Lesser General Public License</span> | 
|  | 37   24 <span class="c"># along with MayaChemTools; if not, see <http://www.gnu.org/licenses/> or</span> | 
|  | 38   25 <span class="c"># write to the Free Software Foundation Inc., 59 Temple Place, Suite 330,</span> | 
|  | 39   26 <span class="c"># Boston, MA, 02111-1307, USA.</span> | 
|  | 40   27 <span class="c">#</span> | 
|  | 41   28 | 
|  | 42   29 <span class="k">use</span> <span class="w">strict</span><span class="sc">;</span> | 
|  | 43   30 <span class="k">use</span> <span class="w">Carp</span><span class="sc">;</span> | 
|  | 44   31 <span class="k">use</span> <span class="w">Text::ParseWords</span><span class="sc">;</span> | 
|  | 45   32 <span class="k">use</span> <span class="w">TextUtil</span><span class="sc">;</span> | 
|  | 46   33 <span class="k">use</span> <span class="w">FileUtil</span><span class="sc">;</span> | 
|  | 47   34 | 
|  | 48   35 <span class="k">use</span> <span class="w">vars</span> <span class="q">qw(@ISA @EXPORT @EXPORT_OK %EXPORT_TAGS)</span><span class="sc">;</span> | 
|  | 49   36 | 
|  | 50   37 <span class="i">@ISA</span> = <span class="q">qw(Exporter)</span><span class="sc">;</span> | 
|  | 51   38 <span class="i">@EXPORT</span> = <span class="q">qw()</span><span class="sc">;</span> | 
|  | 52   39 <span class="i">@EXPORT_OK</span> = <span class="q">qw(GetAminoAcids GetAminoAcidPropertiesData GetAminoAcidPropertiesNames IsAminoAcid IsAminoAcidProperty)</span><span class="sc">;</span> | 
|  | 53   40 | 
|  | 54   41 <span class="i">%EXPORT_TAGS</span> = <span class="s">(</span><span class="w">all</span>  <span class="cm">=></span> <span class="s">[</span><span class="i">@EXPORT</span><span class="cm">,</span> <span class="i">@EXPORT_OK</span><span class="s">]</span><span class="s">)</span><span class="sc">;</span> | 
|  | 55   42 | 
|  | 56   43 <span class="c">#</span> | 
|  | 57   44 <span class="c"># Load amino acids data...</span> | 
|  | 58   45 <span class="c">#</span> | 
|  | 59   46 <span class="k">my</span><span class="s">(</span><span class="i">%AminoAcidDataMap</span><span class="cm">,</span> <span class="i">%AminoAcidThreeLetterCodeMap</span><span class="cm">,</span> <span class="i">%AminoAcidOneLetterCodeMap</span><span class="cm">,</span> <span class="i">%AminoAcidNameMap</span><span class="cm">,</span> <span class="i">@AminoAcidPropertyNames</span><span class="cm">,</span> <span class="i">%AminoAcidPropertyNamesMap</span><span class="cm">,</span> <span class="s">)</span><span class="sc">;</span> | 
|  | 60   47 <span class="i">_LoadAminoAcidsData</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span> | 
|  | 61   48 | 
|  | 62   49 <span class="c">#</span> | 
|  | 63   50 <span class="c"># Get a list of all known amino acids as one of these values:</span> | 
|  | 64   51 <span class="c"># one letter code, three letter code, or amino acid name...</span> | 
|  | 65   52 <span class="c">#</span> | 
|  | 66 <a name="GetAminoAcids-"></a>  53 <span class="k">sub </span><span class="m">GetAminoAcids</span> <span class="s">{</span> | 
|  | 67   54   <span class="k">my</span><span class="s">(</span><span class="i">$NameType</span><span class="cm">,</span> <span class="i">$ThreeLetterCode</span><span class="cm">,</span> <span class="i">$Name</span><span class="cm">,</span> <span class="i">@AminoAcidNames</span><span class="cm">,</span> <span class="i">%AminoAcidNamesMap</span><span class="s">)</span><span class="sc">;</span> | 
|  | 68   55 | 
|  | 69   56   <span class="i">$NameType</span> = <span class="q">'ThreeLetterCode'</span><span class="sc">;</span> | 
|  | 70   57   <span class="k">if</span> <span class="s">(</span><span class="i">@_</span> >= <span class="n">1</span><span class="s">)</span> <span class="s">{</span> | 
|  | 71   58     <span class="s">(</span><span class="i">$NameType</span><span class="s">)</span> = <span class="i">@_</span><span class="sc">;</span> | 
|  | 72   59   <span class="s">}</span> | 
|  | 73   60 | 
|  | 74   61   <span class="c"># Collect names...</span> | 
|  | 75   62   <span class="i">%AminoAcidNamesMap</span> = <span class="s">(</span><span class="s">)</span><span class="sc">;</span> | 
|  | 76   63   <span class="k">for</span> <span class="i">$ThreeLetterCode</span> <span class="s">(</span><span class="k">keys</span> <span class="i">%AminoAcidDataMap</span><span class="s">)</span> <span class="s">{</span> | 
|  | 77   64     <span class="j">NAME :</span> <span class="s">{</span> | 
|  | 78   65       <span class="k">if</span> <span class="s">(</span><span class="i">$NameType</span> =~ <span class="q">/^OneLetterCode$/i</span><span class="s">)</span> <span class="s">{</span><span class="i">$Name</span> = <span class="i">$AminoAcidDataMap</span>{<span class="i">$ThreeLetterCode</span>}{<span class="w">OneLetterCode</span>}<span class="sc">;</span> <span class="k">last</span> <span class="j">NAME</span><span class="sc">;</span> <span class="s">}</span> | 
|  | 79   66       <span class="k">if</span> <span class="s">(</span><span class="i">$NameType</span> =~ <span class="q">/^AminoAcid$/i</span><span class="s">)</span> <span class="s">{</span><span class="i">$Name</span> = <span class="i">$AminoAcidDataMap</span>{<span class="i">$ThreeLetterCode</span>}{<span class="w">AminoAcid</span>}<span class="sc">;</span> <span class="k">last</span> <span class="j">NAME</span><span class="sc">;</span> <span class="s">}</span> | 
|  | 80   67       <span class="i">$Name</span> = <span class="i">$ThreeLetterCode</span><span class="sc">;</span> | 
|  | 81   68     <span class="s">}</span> | 
|  | 82   69     <span class="i">$AminoAcidNamesMap</span>{<span class="i">$Name</span>} = <span class="i">$Name</span><span class="sc">;</span> | 
|  | 83   70   <span class="s">}</span> | 
|  | 84   71 | 
|  | 85   72   <span class="c"># Sort 'em out</span> | 
|  | 86   73   <span class="i">@AminoAcidNames</span> = <span class="s">(</span><span class="s">)</span><span class="sc">;</span> | 
|  | 87   74   <span class="k">for</span> <span class="i">$Name</span> <span class="s">(</span><span class="k">sort</span> <span class="k">keys</span> <span class="i">%AminoAcidNamesMap</span><span class="s">)</span> <span class="s">{</span> | 
|  | 88   75     <span class="k">push</span> <span class="i">@AminoAcidNames</span><span class="cm">,</span> <span class="i">$Name</span><span class="sc">;</span> | 
|  | 89   76   <span class="s">}</span> | 
|  | 90   77 | 
|  | 91   78   <span class="k">return</span> <span class="s">(</span><span class="k">wantarray</span> ? <span class="i">@AminoAcidNames</span> <span class="co">:</span> \<span class="i">@AminoAcidNames</span><span class="s">)</span><span class="sc">;</span> | 
|  | 92   79 <span class="s">}</span> | 
|  | 93   80 | 
|  | 94   81 | 
|  | 95   82 <span class="c">#</span> | 
|  | 96   83 <span class="c"># Get all available properties data for an amino acid using any of these symbols:</span> | 
|  | 97   84 <span class="c"># three letter code; one letter code; name.</span> | 
|  | 98   85 <span class="c">#</span> | 
|  | 99   86 <span class="c"># A reference to a hash array is returned with keys and values representing property</span> | 
|  | 100   87 <span class="c"># name and its values respectively.</span> | 
|  | 101   88 <span class="c">#</span> | 
|  | 102 <a name="GetAminoAcidPropertiesData-"></a>  89 <span class="k">sub </span><span class="m">GetAminoAcidPropertiesData</span> <span class="s">{</span> | 
|  | 103   90   <span class="k">my</span><span class="s">(</span><span class="i">$AminoAcidID</span><span class="s">)</span> = <span class="i">@_</span><span class="sc">;</span> | 
|  | 104   91   <span class="k">my</span><span class="s">(</span><span class="i">$ThreeLetterCode</span><span class="s">)</span><span class="sc">;</span> | 
|  | 105   92 | 
|  | 106   93   <span class="k">if</span> <span class="s">(</span><span class="i">$ThreeLetterCode</span> = <span class="i">_ValidateAminoAcidID</span><span class="s">(</span><span class="i">$AminoAcidID</span><span class="s">)</span><span class="s">)</span> <span class="s">{</span> | 
|  | 107   94     <span class="k">return</span> \<span class="i">%</span>{<span class="i">$AminoAcidDataMap</span>{<span class="i">$ThreeLetterCode</span>}}<span class="sc">;</span> | 
|  | 108   95   <span class="s">}</span> | 
|  | 109   96   <span class="k">else</span> <span class="s">{</span> | 
|  | 110   97     <span class="k">return</span> <span class="k">undef</span><span class="sc">;</span> | 
|  | 111   98   <span class="s">}</span> | 
|  | 112   99 <span class="s">}</span> | 
|  | 113  100 | 
|  | 114  101 <span class="c">#</span> | 
|  | 115  102 <span class="c"># Get names of all available amino acid properties. A reference to  an array containing</span> | 
|  | 116  103 <span class="c"># names of all available properties is returned.</span> | 
|  | 117  104 <span class="c">#</span> | 
|  | 118 <a name="GetAminoAcidPropertiesNames-"></a> 105 <span class="k">sub </span><span class="m">GetAminoAcidPropertiesNames</span> <span class="s">{</span> | 
|  | 119  106   <span class="k">my</span><span class="s">(</span><span class="i">$Mode</span><span class="s">)</span><span class="sc">;</span> | 
|  | 120  107   <span class="k">my</span><span class="s">(</span><span class="i">$PropertyName</span><span class="cm">,</span> <span class="i">@PropertyNames</span><span class="s">)</span><span class="sc">;</span> | 
|  | 121  108 | 
|  | 122  109   <span class="i">$Mode</span> = <span class="q">'ByGroup'</span><span class="sc">;</span> | 
|  | 123  110   <span class="k">if</span> <span class="s">(</span><span class="i">@_</span> == <span class="n">1</span><span class="s">)</span> <span class="s">{</span> | 
|  | 124  111     <span class="s">(</span><span class="i">$Mode</span><span class="s">)</span> = <span class="i">@_</span><span class="sc">;</span> | 
|  | 125  112   <span class="s">}</span> | 
|  | 126  113 | 
|  | 127  114   <span class="i">@PropertyNames</span> = <span class="s">(</span><span class="s">)</span><span class="sc">;</span> | 
|  | 128  115   <span class="k">if</span> <span class="s">(</span><span class="i">$Mode</span> =~ <span class="q">/^Alphabetical$/i</span><span class="s">)</span> <span class="s">{</span> | 
|  | 129  116     <span class="k">my</span><span class="s">(</span><span class="i">$PropertyName</span><span class="s">)</span><span class="sc">;</span> | 
|  | 130  117     <span class="c"># ThreeLetterCode, OneLetterCode, and AminoAcid are always listed first...</span> | 
|  | 131  118     <span class="k">push</span> <span class="i">@PropertyNames</span><span class="cm">,</span> <span class="q">qw(ThreeLetterCode OneLetterCode AminoAcid)</span><span class="sc">;</span> | 
|  | 132  119     <span class="k">for</span> <span class="i">$PropertyName</span> <span class="s">(</span><span class="k">sort</span> <span class="k">keys</span> <span class="i">%AminoAcidPropertyNamesMap</span><span class="s">)</span> <span class="s">{</span> | 
|  | 133  120       <span class="k">if</span> <span class="s">(</span><span class="i">$PropertyName</span> !~ <span class="q">/^(ThreeLetterCode|OneLetterCode|AminoAcid)$/</span><span class="s">)</span> <span class="s">{</span> | 
|  | 134  121         <span class="k">push</span> <span class="i">@PropertyNames</span><span class="cm">,</span> <span class="i">$PropertyName</span><span class="sc">;</span> | 
|  | 135  122       <span class="s">}</span> | 
|  | 136  123     <span class="s">}</span> | 
|  | 137  124   <span class="s">}</span> | 
|  | 138  125   <span class="k">else</span> <span class="s">{</span> | 
|  | 139  126     <span class="k">push</span> <span class="i">@PropertyNames</span><span class="cm">,</span> <span class="i">@AminoAcidPropertyNames</span><span class="sc">;</span> | 
|  | 140  127   <span class="s">}</span> | 
|  | 141  128   <span class="k">return</span> <span class="s">(</span><span class="k">wantarray</span> ? <span class="i">@PropertyNames</span> <span class="co">:</span> \<span class="i">@PropertyNames</span><span class="s">)</span><span class="sc">;</span> | 
|  | 142  129 <span class="s">}</span> | 
|  | 143  130 | 
|  | 144  131 <span class="c">#</span> | 
|  | 145  132 <span class="c"># Is it a known amino acid? Input is either an one/three letter code or a name.</span> | 
|  | 146  133 <span class="c">#</span> | 
|  | 147 <a name="IsAminoAcid-"></a> 134 <span class="k">sub </span><span class="m">IsAminoAcid</span> <span class="s">{</span> | 
|  | 148  135   <span class="k">my</span><span class="s">(</span><span class="i">$AminoAcidID</span><span class="s">)</span> = <span class="i">@_</span><span class="sc">;</span> | 
|  | 149  136   <span class="k">my</span><span class="s">(</span><span class="i">$Status</span><span class="s">)</span><span class="sc">;</span> | 
|  | 150  137 | 
|  | 151  138   <span class="i">$Status</span> = <span class="s">(</span><span class="i">_ValidateAminoAcidID</span><span class="s">(</span><span class="i">$AminoAcidID</span><span class="s">)</span><span class="s">)</span> ? <span class="n">1</span> <span class="co">:</span> <span class="n">0</span><span class="sc">;</span> | 
|  | 152  139 | 
|  | 153  140   <span class="k">return</span> <span class="i">$Status</span><span class="sc">;</span> | 
|  | 154  141 <span class="s">}</span> | 
|  | 155  142 | 
|  | 156  143 | 
|  | 157  144 <span class="c">#</span> | 
|  | 158  145 <span class="c"># Is it an available amino acid property?</span> | 
|  | 159  146 <span class="c">#</span> | 
|  | 160 <a name="IsAminoAcidProperty-"></a> 147 <span class="k">sub </span><span class="m">IsAminoAcidProperty</span> <span class="s">{</span> | 
|  | 161  148   <span class="k">my</span><span class="s">(</span><span class="i">$PropertyName</span><span class="s">)</span> = <span class="i">@_</span><span class="sc">;</span> | 
|  | 162  149   <span class="k">my</span><span class="s">(</span><span class="i">$Status</span><span class="s">)</span><span class="sc">;</span> | 
|  | 163  150 | 
|  | 164  151   <span class="i">$Status</span> = <span class="s">(</span><span class="k">exists</span><span class="s">(</span><span class="i">$AminoAcidPropertyNamesMap</span>{<span class="i">$PropertyName</span>}<span class="s">)</span><span class="s">)</span> ? <span class="n">1</span> <span class="co">:</span> <span class="n">0</span><span class="sc">;</span> | 
|  | 165  152 | 
|  | 166  153   <span class="k">return</span> <span class="i">$Status</span><span class="sc">;</span> | 
|  | 167  154 <span class="s">}</span> | 
|  | 168  155 | 
|  | 169  156 <span class="c">#</span> | 
|  | 170  157 <span class="c"># Implents GetAminoAcid<PropertyName> for a valid proprty name.</span> | 
|  | 171  158 <span class="c">#</span> | 
|  | 172 <a name="AUTOLOAD-"></a> 159 <span class="k">sub </span><span class="m">AUTOLOAD</span> <span class="s">{</span> | 
|  | 173  160   <span class="k">my</span><span class="s">(</span><span class="i">$AminoAcidID</span><span class="s">)</span> = <span class="i">@_</span><span class="sc">;</span> | 
|  | 174  161   <span class="k">my</span><span class="s">(</span><span class="i">$FunctionName</span><span class="cm">,</span> <span class="i">$PropertyName</span><span class="cm">,</span> <span class="i">$PropertyValue</span><span class="cm">,</span> <span class="i">$ThreeLetterCode</span><span class="s">)</span><span class="sc">;</span> | 
|  | 175  162 | 
|  | 176  163   <span class="i">$PropertyValue</span> = <span class="k">undef</span><span class="sc">;</span> | 
|  | 177  164 | 
|  | 178  165   <span class="k">use</span> <span class="w">vars</span> <span class="q">qw($AUTOLOAD)</span><span class="sc">;</span> | 
|  | 179  166   <span class="i">$FunctionName</span> = <span class="i">$AUTOLOAD</span><span class="sc">;</span> | 
|  | 180  167   <span class="i">$FunctionName</span> =~ <span class="q">s/.*:://</span><span class="sc">;</span> | 
|  | 181  168 | 
|  | 182  169   <span class="c"># Only Get<PropertyName> functions are supported...</span> | 
|  | 183  170   <span class="k">if</span> <span class="s">(</span><span class="i">$FunctionName</span> !~ <span class="q">/^Get/</span><span class="s">)</span> <span class="s">{</span> | 
|  | 184  171     <span class="w">croak</span> <span class="q">"Error: Function, AminoAcid::$FunctionName, is not supported by AUTOLOAD in AminoAcid module: Only Get<PropertyName> functions are implemented..."</span><span class="sc">;</span> | 
|  | 185  172   <span class="s">}</span> | 
|  | 186  173 | 
|  | 187  174   <span class="i">$PropertyName</span> = <span class="i">$FunctionName</span><span class="sc">;</span> | 
|  | 188  175   <span class="i">$PropertyName</span> =~  <span class="q">s/^GetAminoAcid//</span><span class="sc">;</span> | 
|  | 189  176   <span class="k">if</span> <span class="s">(</span>!<span class="k">exists</span> <span class="i">$AminoAcidPropertyNamesMap</span>{<span class="i">$PropertyName</span>}<span class="s">)</span> <span class="s">{</span> | 
|  | 190  177     <span class="w">croak</span> <span class="q">"Error: Function, AminoAcid::$FunctionName, is not supported by AUTOLOAD in AminoAcid module: Unknown amino acid property name, $PropertyName, specified..."</span><span class="sc">;</span> | 
|  | 191  178   <span class="s">}</span> | 
|  | 192  179 | 
|  | 193  180   <span class="k">if</span> <span class="s">(</span>!<span class="s">(</span><span class="i">$ThreeLetterCode</span> = <span class="i">_ValidateAminoAcidID</span><span class="s">(</span><span class="i">$AminoAcidID</span><span class="s">)</span><span class="s">)</span><span class="s">)</span> <span class="s">{</span> | 
|  | 194  181     <span class="k">return</span> <span class="k">undef</span><span class="sc">;</span> | 
|  | 195  182   <span class="s">}</span> | 
|  | 196  183   <span class="i">$PropertyValue</span> = <span class="i">$AminoAcidDataMap</span>{<span class="i">$ThreeLetterCode</span>}{<span class="i">$PropertyName</span>}<span class="sc">;</span> | 
|  | 197  184   <span class="k">return</span> <span class="i">$PropertyValue</span><span class="sc">;</span> | 
|  | 198  185 <span class="s">}</span> | 
|  | 199  186 | 
|  | 200  187 | 
|  | 201  188 <span class="c">#</span> | 
|  | 202  189 <span class="c"># Load AminoAcidsData.csv files from <MayaChemTools>/lib directory...</span> | 
|  | 203  190 <span class="c">#</span> | 
|  | 204 <a name="_LoadAminoAcidsData-"></a> 191 <span class="k">sub </span><span class="m">_LoadAminoAcidsData</span> <span class="s">{</span> | 
|  | 205  192   <span class="k">my</span><span class="s">(</span><span class="i">$AminoAcidsDataFile</span><span class="cm">,</span> <span class="i">$MayaChemToolsLibDir</span><span class="s">)</span><span class="sc">;</span> | 
|  | 206  193 | 
|  | 207  194   <span class="i">$MayaChemToolsLibDir</span> = <span class="i">GetMayaChemToolsLibDirName</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span> | 
|  | 208  195 | 
|  | 209  196   <span class="i">$AminoAcidsDataFile</span> =  <span class="q">"$MayaChemToolsLibDir"</span> . <span class="q">"/data/AminoAcidsData.csv"</span><span class="sc">;</span> | 
|  | 210  197 | 
|  | 211  198   <span class="k">if</span> <span class="s">(</span>! <span class="k">-e</span> <span class="q">"$AminoAcidsDataFile"</span><span class="s">)</span> <span class="s">{</span> | 
|  | 212  199     <span class="w">croak</span> <span class="q">"Error: MayaChemTools package file, $AminoAcidsDataFile, is missing: Possible installation problems..."</span><span class="sc">;</span> | 
|  | 213  200   <span class="s">}</span> | 
|  | 214  201 | 
|  | 215  202   <span class="i">_LoadData</span><span class="s">(</span><span class="i">$AminoAcidsDataFile</span><span class="s">)</span><span class="sc">;</span> | 
|  | 216  203 <span class="s">}</span> | 
|  | 217  204 | 
|  | 218  205 <span class="c">#</span> | 
|  | 219  206 <span class="c"># Load AminoAcidsData.csv file from <MayaChemTools>/lib directory...</span> | 
|  | 220  207 <span class="c">#</span> | 
|  | 221 <a name="_LoadData-"></a> 208 <span class="k">sub </span><span class="m">_LoadData</span> <span class="s">{</span> | 
|  | 222  209   <span class="k">my</span><span class="s">(</span><span class="i">$AminoAcidsDataFile</span><span class="s">)</span> = <span class="i">@_</span><span class="sc">;</span> | 
|  | 223  210 | 
|  | 224  211   <span class="i">%AminoAcidDataMap</span> = <span class="s">(</span><span class="s">)</span><span class="sc">;</span> | 
|  | 225  212   <span class="i">@AminoAcidPropertyNames</span> = <span class="s">(</span><span class="s">)</span><span class="sc">;</span> | 
|  | 226  213   <span class="i">%AminoAcidPropertyNamesMap</span> = <span class="s">(</span><span class="s">)</span><span class="sc">;</span> | 
|  | 227  214   <span class="i">%AminoAcidThreeLetterCodeMap</span> = <span class="s">(</span><span class="s">)</span><span class="sc">;</span> | 
|  | 228  215   <span class="i">%AminoAcidOneLetterCodeMap</span> = <span class="s">(</span><span class="s">)</span><span class="sc">;</span> | 
|  | 229  216   <span class="i">%AminoAcidNameMap</span> = <span class="s">(</span><span class="s">)</span><span class="sc">;</span> | 
|  | 230  217 | 
|  | 231  218   <span class="c"># Load property data for all amino acids...</span> | 
|  | 232  219   <span class="c">#</span> | 
|  | 233  220   <span class="c"># File Format:</span> | 
|  | 234  221   <span class="c">#"ThreeLetterCode","OneLetterCode","AminoAcid","AcidicBasic","PolarNonpolar","Charged","Aromatic","HydrophobicHydophilic","IsoelectricPoint","pKCOOH","pKNH3+","MolecularWeight","MolecularWeightMinusH2O(18.01524)","ExactMass","ExactMassMinusH2O(18.01056)","vanderWaalsVolume","%AccessibleResidues","%BuriedResidues","AlphaHelixChouAndFasman","AlphaHelixDeleageAndRoux","AlphaHelixLevitt","AminoAcidsComposition","AminoAcidsCompositionInSwissProt","AntiparallelBetaStrand","AverageAreaBuried","AverageFlexibility","BetaSheetChouAndFasman","BetaSheetDeleageAndRoux","BetaSheetLevitt","BetaTurnChouAndFasman","BetaTurnDeleageAndRoux","BetaTurnLevitt","Bulkiness","CoilDeleageAndRoux","HPLCHFBARetention","HPLCRetentionAtpH2.1","HPLCRetentionAtpH7.4","HPLCTFARetention","HydrophobicityAbrahamAndLeo","HydrophobicityBlack","HydrophobicityBullAndBreese","HydrophobicityChothia","HydrophobicityEisenbergAndOthers","HydrophobicityFauchereAndOthers","HydrophobicityGuy","HydrophobicityHPLCAtpH3.4Cowan","HydrophobicityHPLCAtpH7.5Cowan","HydrophobicityHPLCParkerAndOthers","HydrophobicityHPLCWilsonAndOthers","HydrophobicityHoppAndWoods","HydrophobicityJanin","HydrophobicityKyteAndDoolittle","HydrophobicityManavalanAndOthers","HydrophobicityMiyazawaAndOthers","HydrophobicityOMHSweetAndOthers","HydrophobicityRaoAndArgos","HydrophobicityRfMobility","HydrophobicityRoseAndOthers","HydrophobicityRoseman","HydrophobicityWellingAndOthers","HydrophobicityWolfendenAndOthers","MolecularWeight","NumberOfCodons","ParallelBetaStrand","PolarityGrantham","PolarityZimmerman","RatioHeteroEndToSide","RecognitionFactors","Refractivity","RelativeMutability","TotalBetaStrand","LinearStructure","LinearStructureAtpH7.4"</span> | 
|  | 235  222   <span class="c">#</span> | 
|  | 236  223   <span class="c">#</span> | 
|  | 237  224   <span class="k">my</span><span class="s">(</span><span class="i">$ThreeLetterCode</span><span class="cm">,</span> <span class="i">$OneLetterCode</span><span class="cm">,</span> <span class="i">$AminoAcidName</span><span class="cm">,</span> <span class="i">$Line</span><span class="cm">,</span> <span class="i">$NumOfCols</span><span class="cm">,</span> <span class="i">$InDelim</span><span class="cm">,</span> <span class="i">$Index</span><span class="cm">,</span> <span class="i">$Name</span><span class="cm">,</span> <span class="i">$Value</span><span class="cm">,</span> <span class="i">$Units</span><span class="cm">,</span> <span class="i">@LineWords</span><span class="cm">,</span> <span class="i">@ColLabels</span><span class="s">)</span><span class="sc">;</span> | 
|  | 238  225 | 
|  | 239  226   <span class="i">$InDelim</span> = <span class="q">"\,"</span><span class="sc">;</span> | 
|  | 240  227   <span class="k">open</span> <span class="w">AMINOACIDSDATAFILE</span><span class="cm">,</span> <span class="q">"$AminoAcidsDataFile"</span> <span class="k">or</span> <span class="w">croak</span> <span class="q">"Couldn't open $AminoAcidsDataFile: $! ..."</span><span class="sc">;</span> | 
|  | 241  228 | 
|  | 242  229   <span class="c"># Skip lines up to column labels...</span> | 
|  | 243  230   <span class="j">LINE:</span> <span class="k">while</span> <span class="s">(</span><span class="i">$Line</span> = <span class="i">GetTextLine</span><span class="s">(</span>\<span class="i">*AMINOACIDSDATAFILE</span><span class="s">)</span><span class="s">)</span> <span class="s">{</span> | 
|  | 244  231     <span class="k">if</span> <span class="s">(</span><span class="i">$Line</span> !~ <span class="q">/^#/</span><span class="s">)</span> <span class="s">{</span> | 
|  | 245  232       <span class="k">last</span> <span class="j">LINE</span><span class="sc">;</span> | 
|  | 246  233     <span class="s">}</span> | 
|  | 247  234   <span class="s">}</span> | 
|  | 248  235   <span class="i">@ColLabels</span>= <span class="i">quotewords</span><span class="s">(</span><span class="i">$InDelim</span><span class="cm">,</span> <span class="n">0</span><span class="cm">,</span> <span class="i">$Line</span><span class="s">)</span><span class="sc">;</span> | 
|  | 249  236   <span class="i">$NumOfCols</span> = <span class="i">@ColLabels</span><span class="sc">;</span> | 
|  | 250  237 | 
|  | 251  238   <span class="c"># Extract property names from column labels...</span> | 
|  | 252  239   <span class="i">@AminoAcidPropertyNames</span> = <span class="s">(</span><span class="s">)</span><span class="sc">;</span> | 
|  | 253  240   <span class="k">for</span> <span class="i">$Index</span> <span class="s">(</span><span class="n">0</span> .. <span class="i">$#ColLabels</span><span class="s">)</span> <span class="s">{</span> | 
|  | 254  241     <span class="i">$Name</span> = <span class="i">$ColLabels</span>[<span class="i">$Index</span>]<span class="sc">;</span> | 
|  | 255  242     <span class="k">push</span> <span class="i">@AminoAcidPropertyNames</span><span class="cm">,</span> <span class="i">$Name</span><span class="sc">;</span> | 
|  | 256  243 | 
|  | 257  244     <span class="c"># Store property names...</span> | 
|  | 258  245     <span class="i">$AminoAcidPropertyNamesMap</span>{<span class="i">$Name</span>} = <span class="i">$Name</span><span class="sc">;</span> | 
|  | 259  246   <span class="s">}</span> | 
|  | 260  247 | 
|  | 261  248   <span class="c"># Process amino acid data...</span> | 
|  | 262  249   <span class="j">LINE:</span> <span class="k">while</span> <span class="s">(</span><span class="i">$Line</span> = <span class="i">GetTextLine</span><span class="s">(</span>\<span class="i">*AMINOACIDSDATAFILE</span><span class="s">)</span><span class="s">)</span> <span class="s">{</span> | 
|  | 263  250     <span class="k">if</span> <span class="s">(</span><span class="i">$Line</span> =~ <span class="q">/^#/</span><span class="s">)</span> <span class="s">{</span> | 
|  | 264  251       <span class="k">next</span> <span class="j">LINE</span><span class="sc">;</span> | 
|  | 265  252     <span class="s">}</span> | 
|  | 266  253     <span class="i">@LineWords</span> = <span class="s">(</span><span class="s">)</span><span class="sc">;</span> | 
|  | 267  254     <span class="i">@LineWords</span> = <span class="i">quotewords</span><span class="s">(</span><span class="i">$InDelim</span><span class="cm">,</span> <span class="n">0</span><span class="cm">,</span> <span class="i">$Line</span><span class="s">)</span><span class="sc">;</span> | 
|  | 268  255     <span class="k">if</span> <span class="s">(</span><span class="i">@LineWords</span> != <span class="i">$NumOfCols</span><span class="s">)</span> <span class="s">{</span> | 
|  | 269  256       <span class="w">croak</span> <span class="q">"Error: The number of data fields, @LineWords, in $AminoAcidsDataFile must be $NumOfCols.\nLine: $Line..."</span><span class="sc">;</span> | 
|  | 270  257     <span class="s">}</span> | 
|  | 271  258     <span class="i">$ThreeLetterCode</span> = <span class="i">$LineWords</span>[<span class="n">0</span>]<span class="sc">;</span> <span class="i">$OneLetterCode</span> = <span class="i">$LineWords</span>[<span class="n">1</span>]<span class="sc">;</span> <span class="i">$AminoAcidName</span> = <span class="i">$LineWords</span>[<span class="n">3</span>]<span class="sc">;</span> | 
|  | 272  259     <span class="k">if</span> <span class="s">(</span><span class="k">exists</span> <span class="i">$AminoAcidDataMap</span>{<span class="i">$ThreeLetterCode</span>}<span class="s">)</span> <span class="s">{</span> | 
|  | 273  260       <span class="w">carp</span> <span class="q">"Warning: Ignoring data for amino acid $ThreeLetterCode: It has already been loaded.\nLine: $Line...."</span><span class="sc">;</span> | 
|  | 274  261       <span class="k">next</span> <span class="j">LINE</span><span class="sc">;</span> | 
|  | 275  262     <span class="s">}</span> | 
|  | 276  263 | 
|  | 277  264     <span class="c"># Store all the values...</span> | 
|  | 278  265     <span class="i">%</span>{<span class="i">$AminoAcidDataMap</span>{<span class="i">$ThreeLetterCode</span>}} = <span class="s">(</span><span class="s">)</span><span class="sc">;</span> | 
|  | 279  266     <span class="k">for</span> <span class="i">$Index</span> <span class="s">(</span><span class="n">0</span> .. <span class="i">$#LineWords</span><span class="s">)</span> <span class="s">{</span> | 
|  | 280  267       <span class="i">$Name</span> = <span class="i">$AminoAcidPropertyNames</span>[<span class="i">$Index</span>]<span class="sc">;</span> | 
|  | 281  268       <span class="i">$Value</span> = <span class="i">$LineWords</span>[<span class="i">$Index</span>]<span class="sc">;</span> | 
|  | 282  269       <span class="i">$AminoAcidDataMap</span>{<span class="i">$ThreeLetterCode</span>}{<span class="i">$Name</span>} = <span class="i">$Value</span><span class="sc">;</span> | 
|  | 283  270     <span class="s">}</span> | 
|  | 284  271   <span class="s">}</span> | 
|  | 285  272   <span class="k">close</span> <span class="w">AMINOACIDSDATAFILE</span><span class="sc">;</span> | 
|  | 286  273 | 
|  | 287  274   <span class="c"># Setup one letter and amino acid name maps...</span> | 
|  | 288  275   <span class="i">_SetupAminoAcidIDMap</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span> | 
|  | 289  276 <span class="s">}</span> | 
|  | 290  277 | 
|  | 291  278 | 
|  | 292  279 <span class="c">#</span> | 
|  | 293  280 <span class="c"># Setup lowercase three/one letter code and name maps pointing</span> | 
|  | 294  281 <span class="c"># to three letter code as show in data file.</span> | 
|  | 295  282 <span class="c">#</span> | 
|  | 296 <a name="_SetupAminoAcidIDMap-"></a> 283 <span class="k">sub </span><span class="m">_SetupAminoAcidIDMap</span> <span class="s">{</span> | 
|  | 297  284   <span class="k">my</span><span class="s">(</span><span class="i">$ThreeLetterCode</span><span class="cm">,</span> <span class="i">$OneLetterCode</span><span class="cm">,</span> <span class="i">$AminoAcidName</span><span class="s">)</span><span class="sc">;</span> | 
|  | 298  285 | 
|  | 299  286   <span class="i">%AminoAcidThreeLetterCodeMap</span> = <span class="s">(</span><span class="s">)</span><span class="sc">;</span> | 
|  | 300  287   <span class="i">%AminoAcidOneLetterCodeMap</span> = <span class="s">(</span><span class="s">)</span><span class="sc">;</span> | 
|  | 301  288   <span class="i">%AminoAcidNameMap</span> = <span class="s">(</span><span class="s">)</span><span class="sc">;</span> | 
|  | 302  289 | 
|  | 303  290   <span class="k">for</span> <span class="i">$ThreeLetterCode</span> <span class="s">(</span><span class="k">keys</span> <span class="i">%AminoAcidDataMap</span><span class="s">)</span> <span class="s">{</span> | 
|  | 304  291     <span class="i">$OneLetterCode</span> = <span class="i">$AminoAcidDataMap</span>{<span class="i">$ThreeLetterCode</span>}{<span class="w">OneLetterCode</span>}<span class="sc">;</span> | 
|  | 305  292     <span class="i">$AminoAcidName</span> = <span class="i">$AminoAcidDataMap</span>{<span class="i">$ThreeLetterCode</span>}{<span class="w">AminoAcid</span>}<span class="sc">;</span> | 
|  | 306  293 | 
|  | 307  294     <span class="i">$AminoAcidThreeLetterCodeMap</span>{<span class="k">lc</span><span class="s">(</span><span class="i">$ThreeLetterCode</span><span class="s">)</span>} = <span class="i">$ThreeLetterCode</span><span class="sc">;</span> | 
|  | 308  295     <span class="i">$AminoAcidOneLetterCodeMap</span>{<span class="k">lc</span><span class="s">(</span><span class="i">$OneLetterCode</span><span class="s">)</span>} = <span class="i">$ThreeLetterCode</span><span class="sc">;</span> | 
|  | 309  296     <span class="i">$AminoAcidNameMap</span>{<span class="k">lc</span><span class="s">(</span><span class="i">$AminoAcidName</span><span class="s">)</span>} = <span class="i">$ThreeLetterCode</span><span class="sc">;</span> | 
|  | 310  297   <span class="s">}</span> | 
|  | 311  298 <span class="s">}</span> | 
|  | 312  299 | 
|  | 313  300 <span class="c"># Validate amino acid ID...</span> | 
|  | 314 <a name="_ValidateAminoAcidID-"></a> 301 <span class="k">sub </span><span class="m">_ValidateAminoAcidID</span> <span class="s">{</span> | 
|  | 315  302   <span class="k">my</span><span class="s">(</span><span class="i">$AminoAcidID</span><span class="s">)</span> = <span class="i">@_</span><span class="sc">;</span> | 
|  | 316  303   <span class="k">my</span><span class="s">(</span><span class="i">$ThreeLetterCode</span><span class="s">)</span><span class="sc">;</span> | 
|  | 317  304 | 
|  | 318  305 | 
|  | 319  306   <span class="k">if</span> <span class="s">(</span><span class="k">length</span><span class="s">(</span><span class="i">$AminoAcidID</span><span class="s">)</span> == <span class="n">3</span><span class="s">)</span> <span class="s">{</span> | 
|  | 320  307     <span class="k">if</span> <span class="s">(</span>! <span class="k">exists</span> <span class="i">$AminoAcidThreeLetterCodeMap</span>{<span class="k">lc</span><span class="s">(</span><span class="i">$AminoAcidID</span><span class="s">)</span>}<span class="s">)</span> <span class="s">{</span> | 
|  | 321  308       <span class="k">return</span> <span class="k">undef</span><span class="sc">;</span> | 
|  | 322  309     <span class="s">}</span> | 
|  | 323  310     <span class="i">$ThreeLetterCode</span> = <span class="i">$AminoAcidThreeLetterCodeMap</span>{<span class="k">lc</span><span class="s">(</span><span class="i">$AminoAcidID</span><span class="s">)</span>}<span class="sc">;</span> | 
|  | 324  311   <span class="s">}</span> | 
|  | 325  312   <span class="k">elsif</span> <span class="s">(</span><span class="k">length</span><span class="s">(</span><span class="i">$AminoAcidID</span><span class="s">)</span> == <span class="n">1</span><span class="s">)</span> <span class="s">{</span> | 
|  | 326  313     <span class="k">if</span> <span class="s">(</span>! <span class="k">exists</span> <span class="i">$AminoAcidOneLetterCodeMap</span>{<span class="k">lc</span><span class="s">(</span><span class="i">$AminoAcidID</span><span class="s">)</span>}<span class="s">)</span> <span class="s">{</span> | 
|  | 327  314       <span class="k">return</span> <span class="k">undef</span><span class="sc">;</span> | 
|  | 328  315     <span class="s">}</span> | 
|  | 329  316     <span class="i">$ThreeLetterCode</span> = <span class="i">$AminoAcidOneLetterCodeMap</span>{<span class="k">lc</span><span class="s">(</span><span class="i">$AminoAcidID</span><span class="s">)</span>}<span class="sc">;</span> | 
|  | 330  317   <span class="s">}</span> | 
|  | 331  318   <span class="k">else</span> <span class="s">{</span> | 
|  | 332  319     <span class="k">if</span> <span class="s">(</span>! <span class="k">exists</span> <span class="i">$AminoAcidNameMap</span>{<span class="k">lc</span><span class="s">(</span><span class="i">$AminoAcidID</span><span class="s">)</span>}<span class="s">)</span> <span class="s">{</span> | 
|  | 333  320       <span class="k">return</span> <span class="k">undef</span><span class="sc">;</span> | 
|  | 334  321     <span class="s">}</span> | 
|  | 335  322     <span class="i">$ThreeLetterCode</span> = <span class="i">$AminoAcidNameMap</span>{<span class="k">lc</span><span class="s">(</span><span class="i">$AminoAcidID</span><span class="s">)</span>}<span class="sc">;</span> | 
|  | 336  323   <span class="s">}</span> | 
|  | 337  324   <span class="k">return</span> <span class="i">$ThreeLetterCode</span><span class="sc">;</span> | 
|  | 338  325 <span class="s">}</span> | 
|  | 339  326 | 
|  | 340  327 | 
|  | 341 <a name="EOF-"></a></pre> | 
|  | 342 <p> </p> | 
|  | 343 <br /> | 
|  | 344 <center> | 
|  | 345 <img src="../../../images/h2o2.png"> | 
|  | 346 </center> | 
|  | 347 </body> | 
|  | 348 </html> |