0
|
1 <html>
|
|
2 <head>
|
|
3 <title>MayaChemTools:Code:AnalyzeSequenceFilesData.pl</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 1 #!/usr/bin/perl -w
|
|
15 2 <span class="c">#</span>
|
|
16 3 <span class="c"># $RCSfile: AnalyzeSequenceFilesData.pl,v $</span>
|
|
17 4 <span class="c"># $Date: 2015/02/28 20:46:04 $</span>
|
|
18 5 <span class="c"># $Revision: 1.33 $</span>
|
|
19 6 <span class="c">#</span>
|
|
20 7 <span class="c"># Author: Manish Sud <msud@san.rr.com></span>
|
|
21 8 <span class="c">#</span>
|
|
22 9 <span class="c"># Copyright (C) 2015 Manish Sud. All rights reserved.</span>
|
|
23 10 <span class="c">#</span>
|
|
24 11 <span class="c"># This file is part of MayaChemTools.</span>
|
|
25 12 <span class="c">#</span>
|
|
26 13 <span class="c"># MayaChemTools is free software; you can redistribute it and/or modify it under</span>
|
|
27 14 <span class="c"># the terms of the GNU Lesser General Public License as published by the Free</span>
|
|
28 15 <span class="c"># Software Foundation; either version 3 of the License, or (at your option) any</span>
|
|
29 16 <span class="c"># later version.</span>
|
|
30 17 <span class="c">#</span>
|
|
31 18 <span class="c"># MayaChemTools is distributed in the hope that it will be useful, but without</span>
|
|
32 19 <span class="c"># any warranty; without even the implied warranty of merchantability of fitness</span>
|
|
33 20 <span class="c"># for a particular purpose. See the GNU Lesser General Public License for more</span>
|
|
34 21 <span class="c"># details.</span>
|
|
35 22 <span class="c">#</span>
|
|
36 23 <span class="c"># You should have received a copy of the GNU Lesser General Public License</span>
|
|
37 24 <span class="c"># along with MayaChemTools; if not, see <http://www.gnu.org/licenses/> or</span>
|
|
38 25 <span class="c"># write to the Free Software Foundation Inc., 59 Temple Place, Suite 330,</span>
|
|
39 26 <span class="c"># Boston, MA, 02111-1307, USA.</span>
|
|
40 27 <span class="c">#</span>
|
|
41 28
|
|
42 29 <span class="k">use</span> <span class="w">strict</span><span class="sc">;</span>
|
|
43 30 <span class="k">use</span> <span class="w">FindBin</span><span class="sc">;</span> <span class="k">use</span> <span class="w">lib</span> <span class="q">"$FindBin::Bin/../lib"</span><span class="sc">;</span>
|
|
44 31 <span class="k">use</span> <span class="w">Getopt::Long</span><span class="sc">;</span>
|
|
45 32 <span class="k">use</span> <span class="w">File::Basename</span><span class="sc">;</span>
|
|
46 33 <span class="k">use</span> <span class="w">Text::ParseWords</span><span class="sc">;</span>
|
|
47 34 <span class="k">use</span> <span class="w">Benchmark</span><span class="sc">;</span>
|
|
48 35 <span class="k">use</span> <span class="w">FileUtil</span><span class="sc">;</span>
|
|
49 36 <span class="k">use</span> <span class="w">TextUtil</span><span class="sc">;</span>
|
|
50 37 <span class="k">use</span> <span class="w">SequenceFileUtil</span><span class="sc">;</span>
|
|
51 38 <span class="k">use</span> <span class="w">AminoAcids</span><span class="sc">;</span>
|
|
52 39 <span class="k">use</span> <span class="w">NucleicAcids</span><span class="sc">;</span>
|
|
53 40
|
|
54 41 <span class="k">my</span><span class="s">(</span><span class="i">$ScriptName</span><span class="cm">,</span> <span class="i">%Options</span><span class="cm">,</span> <span class="i">$StartTime</span><span class="cm">,</span> <span class="i">$EndTime</span><span class="cm">,</span> <span class="i">$TotalTime</span><span class="s">)</span><span class="sc">;</span>
|
|
55 42
|
|
56 43 <span class="c"># Autoflush STDOUT</span>
|
|
57 44 <span class="i">$|</span> = <span class="n">1</span><span class="sc">;</span>
|
|
58 45
|
|
59 46 <span class="c"># Starting message...</span>
|
|
60 47 <span class="i">$ScriptName</span> = <span class="i">basename</span><span class="s">(</span><span class="i">$0</span><span class="s">)</span><span class="sc">;</span>
|
|
61 48 <span class="k">print</span> <span class="q">"\n$ScriptName: Starting...\n\n"</span><span class="sc">;</span>
|
|
62 49 <span class="i">$StartTime</span> = <span class="w">new</span> <span class="w">Benchmark</span><span class="sc">;</span>
|
|
63 50
|
|
64 51 <span class="c"># Setup script usage message...</span>
|
|
65 52 <span class="i">SetupScriptUsage</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
66 53 <span class="k">if</span> <span class="s">(</span><span class="i">$Options</span>{<span class="w">help</span>} || <span class="i">@ARGV</span> < <span class="n">1</span><span class="s">)</span> <span class="s">{</span>
|
|
67 54 <span class="k">die</span> <span class="i">GetUsageFromPod</span><span class="s">(</span><span class="q">"$FindBin::Bin/$ScriptName"</span><span class="s">)</span><span class="sc">;</span>
|
|
68 55 <span class="s">}</span>
|
|
69 56
|
|
70 57 <span class="c"># Expand wild card file names...</span>
|
|
71 58 <span class="k">my</span><span class="s">(</span><span class="i">@SequenceFilesList</span><span class="s">)</span><span class="sc">;</span>
|
|
72 59 <span class="i">@SequenceFilesList</span> = <span class="i">ExpandFileNames</span><span class="s">(</span>\<span class="i">@ARGV</span><span class="cm">,</span> <span class="q">"aln msf fasta fta pir"</span><span class="s">)</span><span class="sc">;</span>
|
|
73 60
|
|
74 61 <span class="k">print</span> <span class="q">"Processing options...\n"</span><span class="sc">;</span>
|
|
75 62 <span class="k">my</span><span class="s">(</span><span class="i">%OptionsInfo</span><span class="s">)</span><span class="sc">;</span>
|
|
76 63 <span class="i">ProcessOptions</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
77 64
|
|
78 65 <span class="c"># Set up information about input files...</span>
|
|
79 66 <span class="k">print</span> <span class="q">"Checking input sequence file(s)...\n"</span><span class="sc">;</span>
|
|
80 67 <span class="k">my</span><span class="s">(</span><span class="i">%SequenceFilesInfo</span><span class="s">)</span><span class="sc">;</span>
|
|
81 68 <span class="i">RetrieveSequenceFilesInfo</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
82 69 <span class="i">SetupSequenceRegionsData</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
83 70
|
|
84 71 <span class="c"># Process input files..</span>
|
|
85 72 <span class="k">my</span><span class="s">(</span><span class="i">$FileIndex</span><span class="s">)</span><span class="sc">;</span>
|
|
86 73 <span class="k">if</span> <span class="s">(</span><span class="i">@SequenceFilesList</span> > <span class="n">1</span><span class="s">)</span> <span class="s">{</span>
|
|
87 74 <span class="k">print</span> <span class="q">"\nProcessing sequence files...\n"</span><span class="sc">;</span>
|
|
88 75 <span class="s">}</span>
|
|
89 76 <span class="k">for</span> <span class="i">$FileIndex</span> <span class="s">(</span><span class="n">0</span> .. <span class="i">$#SequenceFilesList</span><span class="s">)</span> <span class="s">{</span>
|
|
90 77 <span class="k">if</span> <span class="s">(</span><span class="i">$SequenceFilesInfo</span>{<span class="w">FilesOkay</span>}[<span class="i">$FileIndex</span>]<span class="s">)</span> <span class="s">{</span>
|
|
91 78 <span class="k">print</span> <span class="q">"\nProcessing file $SequenceFilesList[$FileIndex]...\n"</span><span class="sc">;</span>
|
|
92 79 <span class="i">AnalyzeSequenceFileData</span><span class="s">(</span><span class="i">$FileIndex</span><span class="s">)</span><span class="sc">;</span>
|
|
93 80 <span class="s">}</span>
|
|
94 81 <span class="s">}</span>
|
|
95 82 <span class="k">print</span> <span class="q">"\n$ScriptName:Done...\n\n"</span><span class="sc">;</span>
|
|
96 83
|
|
97 84 <span class="i">$EndTime</span> = <span class="w">new</span> <span class="w">Benchmark</span><span class="sc">;</span>
|
|
98 85 <span class="i">$TotalTime</span> = <span class="w">timediff</span> <span class="s">(</span><span class="i">$EndTime</span><span class="cm">,</span> <span class="i">$StartTime</span><span class="s">)</span><span class="sc">;</span>
|
|
99 86 <span class="k">print</span> <span class="q">"Total time: "</span><span class="cm">,</span> <span class="i">timestr</span><span class="s">(</span><span class="i">$TotalTime</span><span class="s">)</span><span class="cm">,</span> <span class="q">"\n"</span><span class="sc">;</span>
|
|
100 87
|
|
101 88 <span class="c">###############################################################################</span>
|
|
102 89
|
|
103 90 <span class="c"># Analyze sequence file...</span>
|
|
104 <a name="AnalyzeSequenceFileData-"></a> 91 <span class="k">sub </span><span class="m">AnalyzeSequenceFileData</span> <span class="s">{</span>
|
|
105 92 <span class="k">my</span><span class="s">(</span><span class="i">$FileIndex</span><span class="s">)</span> = <span class="i">@_</span><span class="sc">;</span>
|
|
106 93 <span class="k">my</span><span class="s">(</span><span class="i">$SequenceFile</span><span class="cm">,</span> <span class="i">$SequenceDataRef</span><span class="s">)</span><span class="sc">;</span>
|
|
107 94
|
|
108 95 <span class="i">$SequenceFile</span> = <span class="i">$SequenceFilesList</span>[<span class="i">$FileIndex</span>]<span class="sc">;</span>
|
|
109 96
|
|
110 97 <span class="k">open</span> <span class="w">SEQUENCEFILE</span><span class="cm">,</span> <span class="q">"$SequenceFile"</span> <span class="k">or</span> <span class="k">die</span> <span class="q">"Error: Can't open $SequenceFile: $! \n"</span><span class="sc">;</span>
|
|
111 98 <span class="i">$SequenceDataRef</span> = <span class="i">ReadSequenceFile</span><span class="s">(</span><span class="i">$SequenceFile</span><span class="s">)</span><span class="sc">;</span>
|
|
112 99 <span class="k">close</span> <span class="w">SEQUENCEFILE</span><span class="sc">;</span>
|
|
113 100
|
|
114 101 <span class="k">if</span> <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">CalculatePercentIdentityMatrix</span>}<span class="s">)</span> <span class="s">{</span>
|
|
115 102 <span class="i">CalculatePercentIdentityMatrix</span><span class="s">(</span><span class="i">$FileIndex</span><span class="cm">,</span> <span class="i">$SequenceDataRef</span><span class="s">)</span><span class="sc">;</span>
|
|
116 103 <span class="s">}</span>
|
|
117 104 <span class="k">if</span> <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">PerformResidueFrequencyAnalysis</span>}<span class="s">)</span> <span class="s">{</span>
|
|
118 105 <span class="i">PerformResidueFrequencyAnalysis</span><span class="s">(</span><span class="i">$FileIndex</span><span class="cm">,</span> <span class="i">$SequenceDataRef</span><span class="s">)</span><span class="sc">;</span>
|
|
119 106 <span class="s">}</span>
|
|
120 107 <span class="s">}</span>
|
|
121 108
|
|
122 109 <span class="c"># Calculate percent identity matrix...</span>
|
|
123 <a name="CalculatePercentIdentityMatrix-"></a> 110 <span class="k">sub </span><span class="m">CalculatePercentIdentityMatrix</span> <span class="s">{</span>
|
|
124 111 <span class="k">my</span><span class="s">(</span><span class="i">$FileIndex</span><span class="cm">,</span> <span class="i">$SequenceDataRef</span><span class="s">)</span> = <span class="i">@_</span><span class="sc">;</span>
|
|
125 112 <span class="k">my</span><span class="s">(</span><span class="i">$PercentIdentity</span><span class="cm">,</span> <span class="i">$PercentIdentityMatrixFile</span><span class="cm">,</span> <span class="i">$PercentIdentityMatrixRef</span><span class="cm">,</span> <span class="i">$RowID</span><span class="cm">,</span> <span class="i">$ColID</span><span class="cm">,</span> <span class="i">$Line</span><span class="cm">,</span> <span class="i">@LineWords</span><span class="s">)</span><span class="sc">;</span>
|
|
126 113
|
|
127 114 <span class="i">$PercentIdentityMatrixFile</span> = <span class="i">$SequenceFilesInfo</span>{<span class="w">OutFileRoot</span>}[<span class="i">$FileIndex</span>] . <span class="q">'PercentIdentityMatrix.'</span> . <span class="i">$SequenceFilesInfo</span>{<span class="w">OutFileExt</span>}[<span class="i">$FileIndex</span>]<span class="sc">;</span>
|
|
128 115 <span class="i">$PercentIdentityMatrixRef</span> = <span class="i">CalculatePercentSequenceIdentityMatrix</span><span class="s">(</span><span class="i">$SequenceDataRef</span><span class="cm">,</span> <span class="i">$OptionsInfo</span>{<span class="w">IgnoreGaps</span>}<span class="cm">,</span> <span class="i">$OptionsInfo</span>{<span class="w">Precision</span>}<span class="s">)</span><span class="sc">;</span>
|
|
129 116
|
|
130 117 <span class="k">print</span> <span class="q">"Generating percent identity matrix file $PercentIdentityMatrixFile...\n"</span><span class="sc">;</span>
|
|
131 118 <span class="k">open</span> <span class="w">OUTFILE</span><span class="cm">,</span> <span class="q">">$PercentIdentityMatrixFile"</span> <span class="k">or</span> <span class="k">die</span> <span class="q">"Can't open $PercentIdentityMatrixFile: $!\n"</span><span class="sc">;</span>
|
|
132 119
|
|
133 120 <span class="c"># Write out column labels...</span>
|
|
134 121 <span class="i">@LineWords</span> = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
135 122 <span class="k">push</span> <span class="i">@LineWords</span><span class="cm">,</span> <span class="q">''</span><span class="sc">;</span>
|
|
136 123 <span class="k">for</span> <span class="i">$ColID</span> <span class="s">(</span><span class="i">@</span>{<span class="i">$PercentIdentityMatrixRef</span>->{<span class="w">IDs</span>}}<span class="s">)</span> <span class="s">{</span>
|
|
137 124 <span class="k">push</span> <span class="i">@LineWords</span><span class="cm">,</span> <span class="i">$ColID</span><span class="sc">;</span>
|
|
138 125 <span class="s">}</span>
|
|
139 126 <span class="i">$Line</span> = <span class="i">JoinWords</span><span class="s">(</span>\<span class="i">@LineWords</span><span class="cm">,</span> <span class="i">$OptionsInfo</span>{<span class="w">OutDelim</span>}<span class="cm">,</span> <span class="i">$OptionsInfo</span>{<span class="w">OutQuote</span>}<span class="s">)</span><span class="sc">;</span>
|
|
140 127 <span class="k">print</span> <span class="i">OUTFILE</span> <span class="q">"$Line\n"</span><span class="sc">;</span>
|
|
141 128
|
|
142 129 <span class="c"># Write out rows...</span>
|
|
143 130 <span class="k">for</span> <span class="i">$RowID</span> <span class="s">(</span><span class="i">@</span>{<span class="i">$PercentIdentityMatrixRef</span>->{<span class="w">IDs</span>}}<span class="s">)</span> <span class="s">{</span>
|
|
144 131 <span class="i">@LineWords</span> = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
145 132 <span class="k">push</span> <span class="i">@LineWords</span><span class="cm">,</span> <span class="i">$RowID</span><span class="sc">;</span>
|
|
146 133 <span class="k">for</span> <span class="i">$ColID</span> <span class="s">(</span><span class="i">@</span>{<span class="i">$PercentIdentityMatrixRef</span>->{<span class="w">IDs</span>}}<span class="s">)</span> <span class="s">{</span>
|
|
147 134 <span class="i">$PercentIdentity</span> = <span class="i">$PercentIdentityMatrixRef</span>->{<span class="w">PercentIdentity</span>}{<span class="i">$RowID</span>}{<span class="i">$ColID</span>}<span class="sc">;</span>
|
|
148 135 <span class="k">push</span> <span class="i">@LineWords</span><span class="cm">,</span> <span class="i">$PercentIdentity</span><span class="sc">;</span>
|
|
149 136 <span class="s">}</span>
|
|
150 137 <span class="i">$Line</span> = <span class="i">JoinWords</span><span class="s">(</span>\<span class="i">@LineWords</span><span class="cm">,</span> <span class="i">$OptionsInfo</span>{<span class="w">OutDelim</span>}<span class="cm">,</span> <span class="i">$OptionsInfo</span>{<span class="w">OutQuote</span>}<span class="s">)</span><span class="sc">;</span>
|
|
151 138 <span class="k">print</span> <span class="i">OUTFILE</span> <span class="q">"$Line\n"</span><span class="sc">;</span>
|
|
152 139 <span class="s">}</span>
|
|
153 140 <span class="k">close</span> <span class="w">OUTFILE</span><span class="sc">;</span>
|
|
154 141 <span class="s">}</span>
|
|
155 142
|
|
156 143 <span class="c"># Perform frequency analysis...</span>
|
|
157 <a name="PerformResidueFrequencyAnalysis-"></a> 144 <span class="k">sub </span><span class="m">PerformResidueFrequencyAnalysis</span> <span class="s">{</span>
|
|
158 145 <span class="k">my</span><span class="s">(</span><span class="i">$FileIndex</span><span class="cm">,</span> <span class="i">$SequenceDataRef</span><span class="s">)</span> = <span class="i">@_</span><span class="sc">;</span>
|
|
159 146
|
|
160 147 <span class="i">CountResiduesInRegions</span><span class="s">(</span><span class="i">$FileIndex</span><span class="cm">,</span> <span class="i">$SequenceDataRef</span><span class="s">)</span><span class="sc">;</span>
|
|
161 148 <span class="i">CalculatePercentResidueFrequencyInRegions</span><span class="s">(</span><span class="i">$FileIndex</span><span class="cm">,</span> <span class="i">$SequenceDataRef</span><span class="s">)</span><span class="sc">;</span>
|
|
162 149 <span class="i">GeneratePercentResidueFrequencyOutFilesForRegions</span><span class="s">(</span><span class="i">$FileIndex</span><span class="cm">,</span> <span class="i">$SequenceDataRef</span><span class="s">)</span><span class="sc">;</span>
|
|
163 150 <span class="s">}</span>
|
|
164 151
|
|
165 152 <span class="c"># Count residues...</span>
|
|
166 <a name="CountResiduesInRegions-"></a> 153 <span class="k">sub </span><span class="m">CountResiduesInRegions</span> <span class="s">{</span>
|
|
167 154 <span class="k">my</span><span class="s">(</span><span class="i">$FileIndex</span><span class="cm">,</span> <span class="i">$SequenceDataRef</span><span class="s">)</span> = <span class="i">@_</span><span class="sc">;</span>
|
|
168 155
|
|
169 156 <span class="c"># Setup rerfernce sequence data...</span>
|
|
170 157 <span class="k">my</span><span class="s">(</span><span class="i">$RefereceSequenceID</span><span class="cm">,</span> <span class="i">$RefereceSequence</span><span class="s">)</span><span class="sc">;</span>
|
|
171 158 <span class="i">$RefereceSequenceID</span> = <span class="i">$SequenceFilesInfo</span>{<span class="w">RefereceSequenceID</span>}[<span class="i">$FileIndex</span>]<span class="sc">;</span>
|
|
172 159 <span class="i">$RefereceSequence</span> = <span class="i">$SequenceFilesInfo</span>{<span class="w">RefereceSequence</span>}[<span class="i">$FileIndex</span>]<span class="sc">;</span>
|
|
173 160
|
|
174 161 <span class="c"># Count residues...</span>
|
|
175 162 <span class="k">my</span><span class="s">(</span><span class="i">$RegionID</span><span class="cm">,</span> <span class="i">$StartResNum</span><span class="cm">,</span> <span class="i">$EndResNum</span><span class="cm">,</span> <span class="i">$ResNum</span><span class="cm">,</span> <span class="i">$ResIndex</span><span class="cm">,</span> <span class="i">$ID</span><span class="cm">,</span> <span class="i">$Sequence</span><span class="cm">,</span> <span class="i">$Residue</span><span class="s">)</span><span class="sc">;</span>
|
|
176 163 <span class="k">for</span> <span class="i">$RegionID</span> <span class="s">(</span><span class="i">@</span>{<span class="i">$SequenceFilesInfo</span>{<span class="w">RegionsData</span>}[<span class="i">$FileIndex</span>]{<span class="w">RegionIDs</span>}}<span class="s">)</span> <span class="s">{</span>
|
|
177 164 <span class="i">$StartResNum</span> = <span class="i">$SequenceFilesInfo</span>{<span class="w">RegionsData</span>}[<span class="i">$FileIndex</span>]{<span class="i">$RegionID</span>}{<span class="w">StartResNum</span>}<span class="sc">;</span>
|
|
178 165 <span class="i">$EndResNum</span> = <span class="i">$SequenceFilesInfo</span>{<span class="w">RegionsData</span>}[<span class="i">$FileIndex</span>]{<span class="i">$RegionID</span>}{<span class="w">EndResNum</span>}<span class="sc">;</span>
|
|
179 166 <span class="j">RESNUM:</span> <span class="k">for</span> <span class="i">$ResNum</span> <span class="s">(</span><span class="i">$StartResNum</span> .. <span class="i">$EndResNum</span><span class="s">)</span> <span class="s">{</span>
|
|
180 167 <span class="i">$ResIndex</span> = <span class="i">$ResNum</span> - <span class="n">1</span><span class="sc">;</span>
|
|
181 168 <span class="k">if</span> <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">IgnoreGaps</span>} && <span class="i">$SequenceFilesInfo</span>{<span class="w">RefereceSequenceResNums</span>}[<span class="i">$FileIndex</span>]{<span class="w">IsGap</span>}{<span class="i">$ResNum</span>}<span class="s">)</span> <span class="s">{</span>
|
|
182 169 <span class="k">next</span> <span class="j">RESNUM</span><span class="sc">;</span>
|
|
183 170 <span class="s">}</span>
|
|
184 171 <span class="c"># Go over residues in column $ResNum in all the sequences...</span>
|
|
185 172 <span class="j">ID:</span> <span class="k">for</span> <span class="i">$ID</span> <span class="s">(</span><span class="i">@</span>{<span class="i">$SequenceDataRef</span>->{<span class="w">IDs</span>}}<span class="s">)</span> <span class="s">{</span>
|
|
186 173 <span class="i">$Sequence</span> = <span class="i">$SequenceDataRef</span>->{<span class="w">Sequence</span>}{<span class="i">$ID</span>}<span class="sc">;</span>
|
|
187 174 <span class="i">$Residue</span> = <span class="k">substr</span><span class="s">(</span><span class="i">$Sequence</span><span class="cm">,</span> <span class="i">$ResIndex</span><span class="cm">,</span> <span class="n">1</span><span class="s">)</span><span class="sc">;</span>
|
|
188 175 <span class="k">if</span> <span class="s">(</span><span class="i">IsGapResidue</span><span class="s">(</span><span class="i">$Residue</span><span class="s">)</span><span class="s">)</span> <span class="s">{</span>
|
|
189 176 <span class="i">$SequenceFilesInfo</span>{<span class="w">RegionsData</span>}[<span class="i">$FileIndex</span>]{<span class="i">$RegionID</span>}{<span class="w">Count</span>}{<span class="i">$ResNum</span>}{<span class="w">Gap</span>} += <span class="n">1</span><span class="sc">;</span>
|
|
190 177 <span class="s">}</span>
|
|
191 178 <span class="k">else</span> <span class="s">{</span>
|
|
192 179 <span class="k">if</span> <span class="s">(</span><span class="k">exists</span> <span class="i">$SequenceFilesInfo</span>{<span class="w">RegionsData</span>}[<span class="i">$FileIndex</span>]{<span class="i">$RegionID</span>}{<span class="w">Count</span>}{<span class="i">$ResNum</span>}{<span class="i">$Residue</span>}<span class="s">)</span> <span class="s">{</span>
|
|
193 180 <span class="i">$SequenceFilesInfo</span>{<span class="w">RegionsData</span>}[<span class="i">$FileIndex</span>]{<span class="i">$RegionID</span>}{<span class="w">Count</span>}{<span class="i">$ResNum</span>}{<span class="i">$Residue</span>} += <span class="n">1</span><span class="sc">;</span>
|
|
194 181 <span class="s">}</span>
|
|
195 182 <span class="k">else</span> <span class="s">{</span>
|
|
196 183 <span class="c"># Internal error...</span>
|
|
197 184 <span class="k">print</span> <span class="q">"Warning: Ignoring residue $Residue in sequence $ID during ResidueFrequencyAnalysis calculation: Unknown residue...\n"</span><span class="sc">;</span>
|
|
198 185 <span class="s">}</span>
|
|
199 186 <span class="s">}</span>
|
|
200 187 <span class="s">}</span>
|
|
201 188 <span class="s">}</span>
|
|
202 189 <span class="s">}</span>
|
|
203 190 <span class="s">}</span>
|
|
204 191
|
|
205 192 <span class="c"># Calculate percent frequency for various residues in the sequence regions...</span>
|
|
206 <a name="CalculatePercentResidueFrequencyInRegions-"></a> 193 <span class="k">sub </span><span class="m">CalculatePercentResidueFrequencyInRegions</span> <span class="s">{</span>
|
|
207 194 <span class="k">my</span><span class="s">(</span><span class="i">$FileIndex</span><span class="cm">,</span> <span class="i">$SequenceDataRef</span><span class="s">)</span> = <span class="i">@_</span><span class="sc">;</span>
|
|
208 195 <span class="k">my</span><span class="s">(</span><span class="i">$RegionID</span><span class="cm">,</span> <span class="i">$StartResNum</span><span class="cm">,</span> <span class="i">$EndResNum</span><span class="cm">,</span> <span class="i">$ResNum</span><span class="cm">,</span> <span class="i">$Residue</span><span class="cm">,</span> <span class="i">$Count</span><span class="cm">,</span> <span class="i">$PercentCount</span><span class="cm">,</span> <span class="i">$MaxResiduesCount</span><span class="cm">,</span> <span class="i">$Precision</span><span class="s">)</span><span class="sc">;</span>
|
|
209 196
|
|
210 197 <span class="i">$MaxResiduesCount</span> = <span class="i">$SequenceDataRef</span>->{<span class="w">Count</span>}<span class="sc">;</span>
|
|
211 198 <span class="i">$Precision</span> = <span class="i">$OptionsInfo</span>{<span class="w">Precision</span>}<span class="sc">;</span>
|
|
212 199 <span class="k">for</span> <span class="i">$RegionID</span> <span class="s">(</span><span class="i">@</span>{<span class="i">$SequenceFilesInfo</span>{<span class="w">RegionsData</span>}[<span class="i">$FileIndex</span>]{<span class="w">RegionIDs</span>}}<span class="s">)</span> <span class="s">{</span>
|
|
213 200 <span class="i">$StartResNum</span> = <span class="i">$SequenceFilesInfo</span>{<span class="w">RegionsData</span>}[<span class="i">$FileIndex</span>]{<span class="i">$RegionID</span>}{<span class="w">StartResNum</span>}<span class="sc">;</span>
|
|
214 201 <span class="i">$EndResNum</span> = <span class="i">$SequenceFilesInfo</span>{<span class="w">RegionsData</span>}[<span class="i">$FileIndex</span>]{<span class="i">$RegionID</span>}{<span class="w">EndResNum</span>}<span class="sc">;</span>
|
|
215 202 <span class="j">RESNUM:</span> <span class="k">for</span> <span class="i">$ResNum</span> <span class="s">(</span><span class="i">$StartResNum</span> .. <span class="i">$EndResNum</span><span class="s">)</span> <span class="s">{</span>
|
|
216 203 <span class="k">if</span> <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">IgnoreGaps</span>} && <span class="i">$SequenceFilesInfo</span>{<span class="w">RefereceSequenceResNums</span>}[<span class="i">$FileIndex</span>]{<span class="w">IsGap</span>}{<span class="i">$ResNum</span>}<span class="s">)</span> <span class="s">{</span>
|
|
217 204 <span class="k">next</span> <span class="j">RESNUM</span><span class="sc">;</span>
|
|
218 205 <span class="s">}</span>
|
|
219 206 <span class="k">for</span> <span class="i">$Residue</span> <span class="s">(</span><span class="k">keys</span> <span class="i">%</span>{<span class="i">$SequenceFilesInfo</span>{<span class="w">RegionsData</span>}[<span class="i">$FileIndex</span>]{<span class="i">$RegionID</span>}{<span class="w">Count</span>}{<span class="i">$ResNum</span>}}<span class="s">)</span> <span class="s">{</span>
|
|
220 207 <span class="i">$Count</span> = <span class="i">$SequenceFilesInfo</span>{<span class="w">RegionsData</span>}[<span class="i">$FileIndex</span>]{<span class="i">$RegionID</span>}{<span class="w">Count</span>}{<span class="i">$ResNum</span>}{<span class="i">$Residue</span>}<span class="sc">;</span>
|
|
221 208 <span class="i">$PercentCount</span> = <span class="s">(</span><span class="i">$Count</span> / <span class="i">$MaxResiduesCount</span><span class="s">)</span> * <span class="n">100</span><span class="sc">;</span>
|
|
222 209 <span class="i">$PercentCount</span> = <span class="k">sprintf</span><span class="s">(</span><span class="q">"%.${Precision}f"</span><span class="cm">,</span> <span class="i">$PercentCount</span><span class="s">)</span> + <span class="n">0</span><span class="sc">;</span>
|
|
223 210 <span class="i">$SequenceFilesInfo</span>{<span class="w">RegionsData</span>}[<span class="i">$FileIndex</span>]{<span class="i">$RegionID</span>}{<span class="w">PercentCount</span>}{<span class="i">$ResNum</span>}{<span class="i">$Residue</span>} = <span class="i">$PercentCount</span><span class="sc">;</span>
|
|
224 211 <span class="s">}</span>
|
|
225 212 <span class="s">}</span>
|
|
226 213 <span class="s">}</span>
|
|
227 214 <span class="s">}</span>
|
|
228 215
|
|
229 216 <span class="c"># Generate output files...</span>
|
|
230 <a name="GeneratePercentResidueFrequencyOutFilesForRegions-"></a> 217 <span class="k">sub </span><span class="m">GeneratePercentResidueFrequencyOutFilesForRegions</span> <span class="s">{</span>
|
|
231 218 <span class="k">my</span><span class="s">(</span><span class="i">$FileIndex</span><span class="cm">,</span> <span class="i">$SequenceDataRef</span><span class="s">)</span> = <span class="i">@_</span><span class="sc">;</span>
|
|
232 219
|
|
233 220 <span class="c"># Setup rerfernce sequence data...</span>
|
|
234 221 <span class="k">my</span><span class="s">(</span><span class="i">$RefereceSequenceID</span><span class="cm">,</span> <span class="i">$RefereceSequence</span><span class="s">)</span><span class="sc">;</span>
|
|
235 222 <span class="i">$RefereceSequenceID</span> = <span class="i">$SequenceFilesInfo</span>{<span class="w">RefereceSequenceID</span>}[<span class="i">$FileIndex</span>]<span class="sc">;</span>
|
|
236 223 <span class="i">$RefereceSequence</span> = <span class="i">$SequenceFilesInfo</span>{<span class="w">RefereceSequence</span>}[<span class="i">$FileIndex</span>]<span class="sc">;</span>
|
|
237 224
|
|
238 225 <span class="k">my</span><span class="s">(</span><span class="i">$RegionID</span><span class="cm">,</span> <span class="i">$StartResNum</span><span class="cm">,</span> <span class="i">$EndResNum</span><span class="cm">,</span> <span class="i">$ResNum</span><span class="cm">,</span> <span class="i">$Count</span><span class="cm">,</span> <span class="i">$PercentCount</span><span class="cm">,</span> <span class="i">$Residue</span><span class="cm">,</span> <span class="i">$RegionNum</span><span class="cm">,</span> <span class="i">$RegionOutFile</span><span class="cm">,</span> <span class="i">$PercentRegionOutFile</span><span class="cm">,</span> <span class="i">$OutFileRoot</span><span class="cm">,</span> <span class="i">$OutFileExt</span><span class="cm">,</span> <span class="i">$Line</span><span class="cm">,</span> <span class="i">@LineWords</span><span class="cm">,</span> <span class="i">@PercentLineWords</span><span class="s">)</span><span class="sc">;</span>
|
|
239 226
|
|
240 227 <span class="i">$OutFileRoot</span> = <span class="i">$SequenceFilesInfo</span>{<span class="w">OutFileRoot</span>}[<span class="i">$FileIndex</span>]<span class="sc">;</span>
|
|
241 228 <span class="i">$OutFileExt</span> = <span class="i">$SequenceFilesInfo</span>{<span class="w">OutFileExt</span>}[<span class="i">$FileIndex</span>]<span class="sc">;</span>
|
|
242 229 <span class="i">$RegionNum</span> = <span class="n">0</span><span class="sc">;</span>
|
|
243 230 <span class="k">for</span> <span class="i">$RegionID</span> <span class="s">(</span><span class="i">@</span>{<span class="i">$SequenceFilesInfo</span>{<span class="w">RegionsData</span>}[<span class="i">$FileIndex</span>]{<span class="w">RegionIDs</span>}}<span class="s">)</span> <span class="s">{</span>
|
|
244 231 <span class="i">$RegionNum</span>++<span class="sc">;</span>
|
|
245 232 <span class="i">$StartResNum</span> = <span class="i">$SequenceFilesInfo</span>{<span class="w">RegionsData</span>}[<span class="i">$FileIndex</span>]{<span class="i">$RegionID</span>}{<span class="w">StartResNum</span>}<span class="sc">;</span>
|
|
246 233 <span class="i">$EndResNum</span> = <span class="i">$SequenceFilesInfo</span>{<span class="w">RegionsData</span>}[<span class="i">$FileIndex</span>]{<span class="i">$RegionID</span>}{<span class="w">EndResNum</span>}<span class="sc">;</span>
|
|
247 234
|
|
248 235 <span class="i">$RegionOutFile</span> = <span class="q">"${OutFileRoot}ResidueFrequencyAnalysisRegion${RegionNum}.${OutFileExt}"</span><span class="sc">;</span>
|
|
249 236 <span class="i">$PercentRegionOutFile</span> = <span class="q">"${OutFileRoot}PercentResidueFrequencyAnalysisRegion${RegionNum}.${OutFileExt}"</span><span class="sc">;</span>
|
|
250 237
|
|
251 238 <span class="k">print</span> <span class="q">"Generating $RegionOutFile and $PercentRegionOutFile...\n"</span><span class="sc">;</span>
|
|
252 239 <span class="k">open</span> <span class="w">REGIONOUTFILE</span><span class="cm">,</span> <span class="q">">$RegionOutFile"</span> <span class="k">or</span> <span class="k">die</span> <span class="q">"Error: Can't open $RegionOutFile: $! \n"</span><span class="sc">;</span>
|
|
253 240 <span class="k">open</span> <span class="w">PERCENTREGIONOUTFILE</span><span class="cm">,</span> <span class="q">">$PercentRegionOutFile"</span> <span class="k">or</span> <span class="k">die</span> <span class="q">"Error: Can't open $PercentRegionOutFile: $! \n"</span><span class="sc">;</span>
|
|
254 241
|
|
255 242 <span class="c"># Write out reference residue positions as column values....</span>
|
|
256 243 <span class="i">@LineWords</span> = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
257 244 <span class="k">push</span> <span class="i">@LineWords</span><span class="cm">,</span> <span class="q">''</span><span class="sc">;</span>
|
|
258 245 <span class="j">RESNUM:</span> <span class="k">for</span> <span class="i">$ResNum</span> <span class="s">(</span><span class="i">$StartResNum</span> .. <span class="i">$EndResNum</span><span class="s">)</span> <span class="s">{</span>
|
|
259 246 <span class="k">if</span> <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">IgnoreGaps</span>} && <span class="i">$SequenceFilesInfo</span>{<span class="w">RefereceSequenceResNums</span>}[<span class="i">$FileIndex</span>]{<span class="w">IsGap</span>}{<span class="i">$ResNum</span>}<span class="s">)</span> <span class="s">{</span>
|
|
260 247 <span class="k">next</span> <span class="j">RESNUM</span><span class="sc">;</span>
|
|
261 248 <span class="s">}</span>
|
|
262 249 <span class="k">push</span> <span class="i">@LineWords</span><span class="cm">,</span> <span class="i">$ResNum</span><span class="sc">;</span>
|
|
263 250 <span class="s">}</span>
|
|
264 251 <span class="i">$Line</span> = <span class="i">JoinWords</span><span class="s">(</span>\<span class="i">@LineWords</span><span class="cm">,</span> <span class="i">$OptionsInfo</span>{<span class="w">OutDelim</span>}<span class="cm">,</span> <span class="i">$OptionsInfo</span>{<span class="w">OutQuote</span>}<span class="s">)</span><span class="sc">;</span>
|
|
265 252 <span class="k">print</span> <span class="i">REGIONOUTFILE</span> <span class="q">"$Line\n"</span><span class="sc">;</span>
|
|
266 253 <span class="k">print</span> <span class="i">PERCENTREGIONOUTFILE</span> <span class="q">"$Line\n"</span><span class="sc">;</span>
|
|
267 254
|
|
268 255
|
|
269 256 <span class="c"># Write out row data for each residue; Gap residue is written last...</span>
|
|
270 257 <span class="j">RESIDUE:</span> <span class="k">for</span> <span class="i">$Residue</span> <span class="s">(</span><span class="k">sort</span> <span class="i">@</span>{<span class="i">$SequenceFilesInfo</span>{<span class="w">ResidueCodes</span>}[<span class="i">$FileIndex</span>]}<span class="s">)</span> <span class="s">{</span>
|
|
271 258 <span class="k">if</span> <span class="s">(</span><span class="i">$Residue</span> =~ <span class="q">/^Gap$/i</span><span class="s">)</span> <span class="s">{</span>
|
|
272 259 <span class="k">next</span> <span class="j">RESIDUE</span><span class="sc">;</span>
|
|
273 260 <span class="s">}</span>
|
|
274 261 <span class="i">@LineWords</span> = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
275 262 <span class="k">push</span> <span class="i">@LineWords</span><span class="cm">,</span> <span class="i">$Residue</span><span class="sc">;</span>
|
|
276 263 <span class="i">@PercentLineWords</span> = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
277 264 <span class="k">push</span> <span class="i">@PercentLineWords</span><span class="cm">,</span> <span class="i">$Residue</span><span class="sc">;</span>
|
|
278 265
|
|
279 266 <span class="j">RESNUM:</span> <span class="k">for</span> <span class="i">$ResNum</span> <span class="s">(</span><span class="i">$StartResNum</span> .. <span class="i">$EndResNum</span><span class="s">)</span> <span class="s">{</span>
|
|
280 267 <span class="k">if</span> <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">IgnoreGaps</span>} && <span class="i">$SequenceFilesInfo</span>{<span class="w">RefereceSequenceResNums</span>}[<span class="i">$FileIndex</span>]{<span class="w">IsGap</span>}{<span class="i">$ResNum</span>}<span class="s">)</span> <span class="s">{</span>
|
|
281 268 <span class="k">next</span> <span class="j">RESNUM</span><span class="sc">;</span>
|
|
282 269 <span class="s">}</span>
|
|
283 270 <span class="i">$Count</span> = <span class="i">$SequenceFilesInfo</span>{<span class="w">RegionsData</span>}[<span class="i">$FileIndex</span>]{<span class="i">$RegionID</span>}{<span class="w">Count</span>}{<span class="i">$ResNum</span>}{<span class="i">$Residue</span>}<span class="sc">;</span>
|
|
284 271 <span class="k">push</span> <span class="i">@LineWords</span><span class="cm">,</span> <span class="i">$Count</span><span class="sc">;</span>
|
|
285 272 <span class="i">$PercentCount</span> = <span class="i">$SequenceFilesInfo</span>{<span class="w">RegionsData</span>}[<span class="i">$FileIndex</span>]{<span class="i">$RegionID</span>}{<span class="w">PercentCount</span>}{<span class="i">$ResNum</span>}{<span class="i">$Residue</span>}<span class="sc">;</span>
|
|
286 273 <span class="k">push</span> <span class="i">@PercentLineWords</span><span class="cm">,</span> <span class="i">$PercentCount</span><span class="sc">;</span>
|
|
287 274 <span class="s">}</span>
|
|
288 275 <span class="i">$Line</span> = <span class="i">JoinWords</span><span class="s">(</span>\<span class="i">@LineWords</span><span class="cm">,</span> <span class="i">$OptionsInfo</span>{<span class="w">OutDelim</span>}<span class="cm">,</span> <span class="i">$OptionsInfo</span>{<span class="w">OutQuote</span>}<span class="s">)</span><span class="sc">;</span>
|
|
289 276 <span class="k">print</span> <span class="i">REGIONOUTFILE</span> <span class="q">"$Line\n"</span><span class="sc">;</span>
|
|
290 277
|
|
291 278 <span class="i">$Line</span> = <span class="i">JoinWords</span><span class="s">(</span>\<span class="i">@PercentLineWords</span><span class="cm">,</span> <span class="i">$OptionsInfo</span>{<span class="w">OutDelim</span>}<span class="cm">,</span> <span class="i">$OptionsInfo</span>{<span class="w">OutQuote</span>}<span class="s">)</span><span class="sc">;</span>
|
|
292 279 <span class="k">print</span> <span class="i">PERCENTREGIONOUTFILE</span> <span class="q">"$Line\n"</span><span class="sc">;</span>
|
|
293 280 <span class="s">}</span>
|
|
294 281
|
|
295 282 <span class="c"># Write out data for gap...</span>
|
|
296 283 <span class="i">$Residue</span> = <span class="q">'Gap'</span><span class="sc">;</span>
|
|
297 284 <span class="i">@LineWords</span> = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
298 285 <span class="k">push</span> <span class="i">@LineWords</span><span class="cm">,</span> <span class="i">$Residue</span><span class="sc">;</span>
|
|
299 286 <span class="i">@PercentLineWords</span> = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
300 287 <span class="k">push</span> <span class="i">@PercentLineWords</span><span class="cm">,</span> <span class="i">$Residue</span><span class="sc">;</span>
|
|
301 288
|
|
302 289 <span class="j">RESNUM:</span> <span class="k">for</span> <span class="i">$ResNum</span> <span class="s">(</span><span class="i">$StartResNum</span> .. <span class="i">$EndResNum</span><span class="s">)</span> <span class="s">{</span>
|
|
303 290 <span class="k">if</span> <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">IgnoreGaps</span>} && <span class="i">$SequenceFilesInfo</span>{<span class="w">RefereceSequenceResNums</span>}[<span class="i">$FileIndex</span>]{<span class="w">IsGap</span>}{<span class="i">$ResNum</span>}<span class="s">)</span> <span class="s">{</span>
|
|
304 291 <span class="k">next</span> <span class="j">RESNUM</span><span class="sc">;</span>
|
|
305 292 <span class="s">}</span>
|
|
306 293 <span class="i">$Count</span> = <span class="i">$SequenceFilesInfo</span>{<span class="w">RegionsData</span>}[<span class="i">$FileIndex</span>]{<span class="i">$RegionID</span>}{<span class="w">Count</span>}{<span class="i">$ResNum</span>}{<span class="i">$Residue</span>}<span class="sc">;</span>
|
|
307 294 <span class="k">push</span> <span class="i">@LineWords</span><span class="cm">,</span> <span class="i">$Count</span><span class="sc">;</span>
|
|
308 295
|
|
309 296 <span class="i">$PercentCount</span> = <span class="i">$SequenceFilesInfo</span>{<span class="w">RegionsData</span>}[<span class="i">$FileIndex</span>]{<span class="i">$RegionID</span>}{<span class="w">PercentCount</span>}{<span class="i">$ResNum</span>}{<span class="i">$Residue</span>}<span class="sc">;</span>
|
|
310 297 <span class="k">push</span> <span class="i">@PercentLineWords</span><span class="cm">,</span> <span class="i">$PercentCount</span><span class="sc">;</span>
|
|
311 298 <span class="s">}</span>
|
|
312 299 <span class="i">$Line</span> = <span class="i">JoinWords</span><span class="s">(</span>\<span class="i">@LineWords</span><span class="cm">,</span> <span class="i">$OptionsInfo</span>{<span class="w">OutDelim</span>}<span class="cm">,</span> <span class="i">$OptionsInfo</span>{<span class="w">OutQuote</span>}<span class="s">)</span><span class="sc">;</span>
|
|
313 300 <span class="k">print</span> <span class="i">REGIONOUTFILE</span> <span class="q">"$Line\n"</span><span class="sc">;</span>
|
|
314 301
|
|
315 302 <span class="i">$Line</span> = <span class="i">JoinWords</span><span class="s">(</span>\<span class="i">@PercentLineWords</span><span class="cm">,</span> <span class="i">$OptionsInfo</span>{<span class="w">OutDelim</span>}<span class="cm">,</span> <span class="i">$OptionsInfo</span>{<span class="w">OutQuote</span>}<span class="s">)</span><span class="sc">;</span>
|
|
316 303 <span class="k">print</span> <span class="i">PERCENTREGIONOUTFILE</span> <span class="q">"$Line\n"</span><span class="sc">;</span>
|
|
317 304
|
|
318 305 <span class="k">close</span> <span class="w">REGIONOUTFILE</span><span class="sc">;</span>
|
|
319 306 <span class="k">close</span> <span class="w">PERCENTREGIONOUTFILE</span><span class="sc">;</span>
|
|
320 307 <span class="s">}</span>
|
|
321 308 <span class="s">}</span>
|
|
322 309
|
|
323 310 <span class="c"># Process option values...</span>
|
|
324 <a name="ProcessOptions-"></a> 311 <span class="k">sub </span><span class="m">ProcessOptions</span> <span class="s">{</span>
|
|
325 312 <span class="i">%OptionsInfo</span> = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
326 313
|
|
327 314 <span class="c"># Setup analysis mode...</span>
|
|
328 315 <span class="i">$OptionsInfo</span>{<span class="w">CalculatePercentIdentityMatrix</span>} = <span class="s">(</span><span class="i">$Options</span>{<span class="w">mode</span>} =~ <span class="q">/^(PercentIdentityMatrix|All)$/i</span><span class="s">)</span> ? <span class="n">1</span> <span class="co">:</span> <span class="n">0</span><span class="sc">;</span>
|
|
329 316 <span class="i">$OptionsInfo</span>{<span class="w">PerformResidueFrequencyAnalysis</span>} = <span class="s">(</span><span class="i">$Options</span>{<span class="w">mode</span>} =~ <span class="q">/^(ResidueFrequencyAnalysis|All)$/i</span><span class="s">)</span> ? <span class="n">1</span> <span class="co">:</span> <span class="n">0</span><span class="sc">;</span>
|
|
330 317
|
|
331 318 <span class="c"># Setup delimiter and quotes...</span>
|
|
332 319 <span class="i">$OptionsInfo</span>{<span class="w">OutDelim</span>} = <span class="s">(</span><span class="i">$Options</span>{<span class="w">outdelim</span>} =~ <span class="q">/tab/i</span> <span class="s">)</span> ? <span class="q">"\t"</span> <span class="co">:</span> <span class="s">(</span><span class="s">(</span><span class="i">$Options</span>{<span class="w">outdelim</span>} =~ <span class="q">/semicolon/i</span><span class="s">)</span> ? <span class="q">"\;"</span> <span class="co">:</span> <span class="q">"\,"</span><span class="s">)</span><span class="sc">;</span>
|
|
333 320 <span class="i">$OptionsInfo</span>{<span class="w">OutQuote</span>} = <span class="s">(</span><span class="i">$Options</span>{<span class="w">quote</span>} =~ <span class="q">/yes/i</span> <span class="s">)</span> ? <span class="n">1</span> <span class="co">:</span> <span class="n">0</span><span class="sc">;</span>
|
|
334 321
|
|
335 322 <span class="c"># Setup reference sequence and regions for residue frequence analysis...</span>
|
|
336 323 <span class="i">$OptionsInfo</span>{<span class="w">SpecifiedRefereceSequence</span>} = <span class="i">$Options</span>{<span class="w">referencesequence</span>}<span class="sc">;</span>
|
|
337 324 <span class="i">$OptionsInfo</span>{<span class="w">SpecifiedRegion</span>} = <span class="i">$Options</span>{<span class="w">region</span>}<span class="sc">;</span>
|
|
338 325 <span class="i">@</span>{<span class="i">$OptionsInfo</span>{<span class="w">SpecifiedRegions</span>}} = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
339 326
|
|
340 327 <span class="k">my</span><span class="s">(</span><span class="i">@SpecifiedRegions</span><span class="s">)</span><span class="sc">;</span>
|
|
341 328 <span class="i">@SpecifiedRegions</span> = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
342 329 <span class="k">if</span> <span class="s">(</span><span class="i">$Options</span>{<span class="w">region</span>} =~ <span class="q">/\,/</span><span class="s">)</span> <span class="s">{</span>
|
|
343 330 <span class="i">@SpecifiedRegions</span> = <span class="k">split</span> <span class="q">/\,/</span><span class="cm">,</span> <span class="i">$OptionsInfo</span>{<span class="w">SpecifiedRegion</span>}<span class="sc">;</span>
|
|
344 331 <span class="k">if</span> <span class="s">(</span><span class="i">@SpecifiedRegions</span> % <span class="n">2</span><span class="s">)</span> <span class="s">{</span>
|
|
345 332 <span class="k">die</span> <span class="q">"Error: The value specified, $Options{region}, for option \"--region\" is not valid. Allowed values: \"StartResNum,EndResNum,[StartResNum,EndResNum...]\" or UseCompleteSequence\n"</span><span class="sc">;</span>
|
|
346 333 <span class="s">}</span>
|
|
347 334 <span class="c"># Make sure EndResNum > StartResNum...</span>
|
|
348 335 <span class="k">my</span><span class="s">(</span><span class="i">$StartResNum</span><span class="cm">,</span> <span class="i">$EndResNum</span><span class="cm">,</span> <span class="i">$Index</span><span class="cm">,</span> <span class="i">$RegionNum</span><span class="s">)</span><span class="sc">;</span>
|
|
349 336 <span class="i">$RegionNum</span> = <span class="n">0</span><span class="sc">;</span>
|
|
350 337 <span class="k">for</span> <span class="s">(</span><span class="i">$Index</span> = <span class="n">0</span><span class="sc">;</span> <span class="i">$Index</span> <= <span class="i">$#SpecifiedRegions</span><span class="sc">;</span> <span class="i">$Index</span> += <span class="n">2</span><span class="s">)</span> <span class="s">{</span>
|
|
351 338 <span class="i">$StartResNum</span> = <span class="i">$SpecifiedRegions</span>[<span class="i">$Index</span>]<span class="sc">;</span>
|
|
352 339 <span class="i">$EndResNum</span> = <span class="i">$SpecifiedRegions</span>[<span class="i">$Index</span> + <span class="n">1</span>]<span class="sc">;</span>
|
|
353 340 <span class="i">$RegionNum</span>++<span class="sc">;</span>
|
|
354 341 <span class="k">if</span> <span class="s">(</span>!<span class="i">IsPositiveInteger</span><span class="s">(</span><span class="i">$StartResNum</span><span class="s">)</span><span class="s">)</span> <span class="s">{</span>
|
|
355 342 <span class="k">die</span> <span class="q">"Error: The value specified, $Options{region}, for option \"--region\" is not valid: The start residue number, $StartResNum, must be a positive integer for region $RegionNum.\n"</span><span class="sc">;</span>
|
|
356 343 <span class="s">}</span>
|
|
357 344 <span class="k">if</span> <span class="s">(</span>!<span class="i">IsPositiveInteger</span><span class="s">(</span><span class="i">$EndResNum</span><span class="s">)</span><span class="s">)</span> <span class="s">{</span>
|
|
358 345 <span class="k">die</span> <span class="q">"Error: The value specified, $Options{region}, for option \"--region\" is not valid: The start residue number, $EndResNum, must be a positive integer for region $RegionNum.\n"</span><span class="sc">;</span>
|
|
359 346 <span class="s">}</span>
|
|
360 347 <span class="k">if</span> <span class="s">(</span><span class="i">$StartResNum</span> >= <span class="i">$EndResNum</span><span class="s">)</span> <span class="s">{</span>
|
|
361 348 <span class="k">die</span> <span class="q">"Error: The value specified, $Options{region}, for option \"--region\" is not valid: The start residue number, $StartResNum, must be smaller than end residue number, $EndResNum, for region $RegionNum.\n"</span><span class="sc">;</span>
|
|
362 349 <span class="s">}</span>
|
|
363 350 <span class="s">}</span>
|
|
364 351 <span class="s">}</span>
|
|
365 352 <span class="k">else</span> <span class="s">{</span>
|
|
366 353 <span class="k">if</span> <span class="s">(</span><span class="i">$Options</span>{<span class="w">region</span>} !~ <span class="q">/^UseCompleteSequence$/i</span><span class="s">)</span> <span class="s">{</span>
|
|
367 354 <span class="k">die</span> <span class="q">"Error: The value specified, $Options{region}, for option \"--region\" is not valid. Allowed values: \"StartResNum,EndResNum,[StartResNum,EndResNum...]\" or UseCompleteSequence\n"</span><span class="sc">;</span>
|
|
368 355 <span class="s">}</span>
|
|
369 356 <span class="s">}</span>
|
|
370 357 <span class="k">push</span> <span class="i">@</span>{<span class="i">$OptionsInfo</span>{<span class="w">SpecifiedRegions</span>}}<span class="cm">,</span> <span class="i">@SpecifiedRegions</span><span class="sc">;</span>
|
|
371 358
|
|
372 359 <span class="c"># Miscellaneous options...</span>
|
|
373 360 <span class="i">$OptionsInfo</span>{<span class="w">Precision</span>} = <span class="i">$Options</span>{<span class="w">precision</span>}<span class="sc">;</span>
|
|
374 361 <span class="i">$OptionsInfo</span>{<span class="w">IgnoreGaps</span>} = <span class="s">(</span><span class="i">$Options</span>{<span class="w">ignoregaps</span>} =~ <span class="q">/Yes/i</span><span class="s">)</span> ? <span class="n">1</span> <span class="co">:</span> <span class="n">0</span><span class="sc">;</span>
|
|
375 362 <span class="i">$OptionsInfo</span>{<span class="w">RegionResiduesMode</span>} = <span class="i">$Options</span>{<span class="w">regionresiduesmode</span>}<span class="sc">;</span>
|
|
376 363
|
|
377 364 <span class="i">$OptionsInfo</span>{<span class="w">OverwriteFiles</span>} = <span class="i">$Options</span>{<span class="w">overwrite</span>} ? <span class="n">1</span> <span class="co">:</span> <span class="n">0</span><span class="sc">;</span>
|
|
378 365 <span class="i">$OptionsInfo</span>{<span class="w">OutFileRoot</span>} = <span class="i">$Options</span>{<span class="w">root</span>} ? <span class="i">$Options</span>{<span class="w">root</span>} <span class="co">:</span> <span class="n">0</span><span class="sc">;</span>
|
|
379 366 <span class="s">}</span>
|
|
380 367
|
|
381 368 <span class="c"># Retrieve information about sequence files...</span>
|
|
382 <a name="RetrieveSequenceFilesInfo-"></a> 369 <span class="k">sub </span><span class="m">RetrieveSequenceFilesInfo</span> <span class="s">{</span>
|
|
383 370 <span class="k">my</span><span class="s">(</span><span class="i">$Index</span><span class="cm">,</span> <span class="i">$SequenceFile</span><span class="cm">,</span> <span class="i">$FileSupported</span><span class="cm">,</span> <span class="i">$FileFormat</span><span class="cm">,</span> <span class="i">$SequenceCount</span><span class="cm">,</span> <span class="i">$RefereceSequence</span><span class="cm">,</span> <span class="i">$RefereceSequenceID</span><span class="cm">,</span> <span class="i">$RefereceSequenceLen</span><span class="cm">,</span> <span class="i">$RefereceSequenceWithNoGaps</span><span class="cm">,</span> <span class="i">$RefereceSequenceWithNoGapsLen</span><span class="cm">,</span> <span class="i">$RefereceSequenceRegionCount</span><span class="cm">,</span> <span class="i">$FileDir</span><span class="cm">,</span> <span class="i">$FileName</span><span class="cm">,</span> <span class="i">$FileExt</span><span class="cm">,</span> <span class="i">$OutFileRoot</span><span class="cm">,</span> <span class="i">$OutFileExt</span><span class="cm">,</span> <span class="i">$SequenceDataRef</span><span class="cm">,</span> <span class="i">$SpecifiedRefereceSequence</span><span class="cm">,</span> <span class="i">@SpecifiedRegions</span><span class="cm">,</span> <span class="i">@RefereceSequenceRegions</span><span class="s">)</span><span class="sc">;</span>
|
|
384 371
|
|
385 372 <span class="i">%SequenceFilesInfo</span> = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
386 373 <span class="i">@</span>{<span class="i">$SequenceFilesInfo</span>{<span class="w">FilesOkay</span>}} = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
387 374 <span class="i">@</span>{<span class="i">$SequenceFilesInfo</span>{<span class="w">OutFileRoot</span>}} = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
388 375 <span class="i">@</span>{<span class="i">$SequenceFilesInfo</span>{<span class="w">OutFileExt</span>}} = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
389 376 <span class="i">@</span>{<span class="i">$SequenceFilesInfo</span>{<span class="w">Format</span>}} = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
390 377 <span class="i">@</span>{<span class="i">$SequenceFilesInfo</span>{<span class="w">SequenceCount</span>}} = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
391 378 <span class="i">@</span>{<span class="i">$SequenceFilesInfo</span>{<span class="w">RefereceSequenceID</span>}} = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
392 379 <span class="i">@</span>{<span class="i">$SequenceFilesInfo</span>{<span class="w">RefereceSequence</span>}} = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
393 380 <span class="i">@</span>{<span class="i">$SequenceFilesInfo</span>{<span class="w">RefereceSequenceLen</span>}} = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
394 381 <span class="i">@</span>{<span class="i">$SequenceFilesInfo</span>{<span class="w">RefereceSequenceWithNoGaps</span>}} = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
395 382 <span class="i">@</span>{<span class="i">$SequenceFilesInfo</span>{<span class="w">RefereceSequenceWithNoGapsLen</span>}} = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
396 383 <span class="i">@</span>{<span class="i">$SequenceFilesInfo</span>{<span class="w">RefereceSequenceRegions</span>}} = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
397 384 <span class="i">@</span>{<span class="i">$SequenceFilesInfo</span>{<span class="w">RefereceSequenceRegionCount</span>}} = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
398 385 <span class="i">@</span>{<span class="i">$SequenceFilesInfo</span>{<span class="w">ResidueCodes</span>}} = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
399 386
|
|
400 387 <span class="j">FILELIST:</span> <span class="k">for</span> <span class="i">$Index</span> <span class="s">(</span><span class="n">0</span> .. <span class="i">$#SequenceFilesList</span><span class="s">)</span> <span class="s">{</span>
|
|
401 388 <span class="i">$SequenceFile</span> = <span class="i">$SequenceFilesList</span>[<span class="i">$Index</span>]<span class="sc">;</span>
|
|
402 389 <span class="i">$SequenceFilesInfo</span>{<span class="w">FilesOkay</span>}[<span class="i">$Index</span>] = <span class="n">0</span><span class="sc">;</span>
|
|
403 390 <span class="i">$SequenceFilesInfo</span>{<span class="w">OutFileRoot</span>}[<span class="i">$Index</span>] = <span class="q">''</span><span class="sc">;</span>
|
|
404 391 <span class="i">$SequenceFilesInfo</span>{<span class="w">OutFileExt</span>}[<span class="i">$Index</span>] = <span class="q">''</span><span class="sc">;</span>
|
|
405 392 <span class="i">$SequenceFilesInfo</span>{<span class="w">Format</span>}[<span class="i">$Index</span>] = <span class="q">'NotSupported'</span><span class="sc">;</span>
|
|
406 393 <span class="i">$SequenceFilesInfo</span>{<span class="w">SequenceCount</span>}[<span class="i">$Index</span>] = <span class="n">0</span><span class="sc">;</span>
|
|
407 394 <span class="i">$SequenceFilesInfo</span>{<span class="w">RefereceSequenceID</span>}[<span class="i">$Index</span>] = <span class="q">''</span><span class="sc">;</span>
|
|
408 395 <span class="i">$SequenceFilesInfo</span>{<span class="w">RefereceSequence</span>}[<span class="i">$Index</span>] = <span class="q">''</span><span class="sc">;</span>
|
|
409 396 <span class="i">$SequenceFilesInfo</span>{<span class="w">RefereceSequenceLen</span>}[<span class="i">$Index</span>] = <span class="q">''</span><span class="sc">;</span>
|
|
410 397 <span class="i">$SequenceFilesInfo</span>{<span class="w">RefereceSequenceWithNoGaps</span>}[<span class="i">$Index</span>] = <span class="q">''</span><span class="sc">;</span>
|
|
411 398 <span class="i">$SequenceFilesInfo</span>{<span class="w">RefereceSequenceWithNoGapsLen</span>}[<span class="i">$Index</span>] = <span class="q">''</span><span class="sc">;</span>
|
|
412 399 <span class="i">@</span>{<span class="i">$SequenceFilesInfo</span>{<span class="w">RefereceSequenceRegions</span>}[<span class="i">$Index</span>]} = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
413 400 <span class="i">$SequenceFilesInfo</span>{<span class="w">RefereceSequenceRegionCount</span>}[<span class="i">$Index</span>] = <span class="n">0</span><span class="sc">;</span>
|
|
414 401 <span class="i">@</span>{<span class="i">$SequenceFilesInfo</span>{<span class="w">ResidueCodes</span>}[<span class="i">$Index</span>]} = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
415 402
|
|
416 403 <span class="k">if</span> <span class="s">(</span>! <span class="k">open</span> <span class="w">SEQUENCEFILE</span><span class="cm">,</span> <span class="q">"$SequenceFile"</span><span class="s">)</span> <span class="s">{</span>
|
|
417 404 <span class="k">warn</span> <span class="q">"Warning: Ignoring file $SequenceFile: Couldn't open it: $! \n"</span><span class="sc">;</span>
|
|
418 405 <span class="k">next</span> <span class="j">FILELIST</span><span class="sc">;</span>
|
|
419 406 <span class="s">}</span>
|
|
420 407 <span class="k">close</span> <span class="w">SEQUENCEFILE</span><span class="sc">;</span>
|
|
421 408
|
|
422 409 <span class="s">(</span><span class="i">$FileSupported</span><span class="cm">,</span> <span class="i">$FileFormat</span><span class="s">)</span> = <span class="i">IsSupportedSequenceFile</span><span class="s">(</span><span class="i">$SequenceFile</span><span class="s">)</span><span class="sc">;</span>
|
|
423 410 <span class="k">if</span> <span class="s">(</span>!<span class="i">$FileSupported</span><span class="s">)</span> <span class="s">{</span>
|
|
424 411 <span class="k">warn</span> <span class="q">"Warning: Ignoring file $SequenceFile: Sequence file format is not supported.\n"</span><span class="sc">;</span>
|
|
425 412 <span class="k">next</span> <span class="j">FILELIST</span><span class="sc">;</span>
|
|
426 413 <span class="s">}</span>
|
|
427 414
|
|
428 415 <span class="i">$SequenceDataRef</span> = <span class="i">ReadSequenceFile</span><span class="s">(</span><span class="i">$SequenceFile</span><span class="s">)</span><span class="sc">;</span>
|
|
429 416
|
|
430 417 <span class="i">$SequenceCount</span> = <span class="i">$SequenceDataRef</span>->{<span class="w">Count</span>}<span class="sc">;</span>
|
|
431 418 <span class="k">if</span> <span class="s">(</span>!<span class="i">$SequenceCount</span><span class="s">)</span> <span class="s">{</span>
|
|
432 419 <span class="k">warn</span> <span class="q">"Warning: Ignoring file $SequenceFile: Sequence data is missing.\n"</span><span class="sc">;</span>
|
|
433 420 <span class="k">next</span> <span class="j">FILELIST</span><span class="sc">;</span>
|
|
434 421 <span class="s">}</span>
|
|
435 422
|
|
436 423 <span class="c"># Make sure all sequence lengths are identical...</span>
|
|
437 424 <span class="k">if</span> <span class="s">(</span>!<span class="i">AreSequenceLengthsIdentical</span><span class="s">(</span><span class="i">$SequenceDataRef</span><span class="s">)</span><span class="s">)</span> <span class="s">{</span>
|
|
438 425 <span class="k">warn</span> <span class="q">"Warning: Ignoring file $SequenceFile: Sequence legths are not identical.\n"</span><span class="sc">;</span>
|
|
439 426 <span class="k">next</span> <span class="j">FILELIST</span><span class="sc">;</span>
|
|
440 427 <span class="s">}</span>
|
|
441 428 <span class="i">$SpecifiedRefereceSequence</span> = <span class="i">$OptionsInfo</span>{<span class="w">SpecifiedRefereceSequence</span>}<span class="sc">;</span>
|
|
442 429 <span class="c"># Make sure reference sequence ID is valid...</span>
|
|
443 430 <span class="k">if</span> <span class="s">(</span><span class="i">$SpecifiedRefereceSequence</span> =~ <span class="q">/^UseFirstSequenceID$/i</span><span class="s">)</span> <span class="s">{</span>
|
|
444 431 <span class="i">$RefereceSequenceID</span> = <span class="i">$SequenceDataRef</span>->{<span class="w">IDs</span>}[<span class="n">0</span>]<span class="sc">;</span>
|
|
445 432 <span class="s">}</span>
|
|
446 433 <span class="k">else</span> <span class="s">{</span>
|
|
447 434 <span class="k">if</span> <span class="s">(</span>!<span class="k">exists</span><span class="s">(</span><span class="i">$SequenceDataRef</span>->{<span class="w">Sequence</span>}{<span class="i">$SpecifiedRefereceSequence</span>}<span class="s">)</span><span class="s">)</span> <span class="s">{</span>
|
|
448 435 <span class="k">warn</span> <span class="q">"Warning: Ignoring file $SequenceFile: Rreference sequence ID, $SpecifiedRefereceSequence, specified using option \"--referencesequence\" is missing.\n"</span><span class="sc">;</span>
|
|
449 436 <span class="k">next</span> <span class="j">FILELIST</span><span class="sc">;</span>
|
|
450 437 <span class="s">}</span>
|
|
451 438 <span class="i">$RefereceSequenceID</span> = <span class="i">$SpecifiedRefereceSequence</span><span class="sc">;</span>
|
|
452 439 <span class="s">}</span>
|
|
453 440
|
|
454 441 <span class="c"># Make sure sequence regions corresponding to reference sequence are valid...</span>
|
|
455 442 <span class="i">@RefereceSequenceRegions</span> = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
456 443 <span class="i">$RefereceSequenceRegionCount</span> = <span class="n">0</span><span class="sc">;</span>
|
|
457 444 <span class="i">$RefereceSequence</span> = <span class="i">$SequenceDataRef</span>->{<span class="w">Sequence</span>}{<span class="i">$RefereceSequenceID</span>}<span class="sc">;</span>
|
|
458 445 <span class="i">$RefereceSequenceLen</span> = <span class="k">length</span> <span class="i">$RefereceSequence</span><span class="sc">;</span>
|
|
459 446
|
|
460 447 <span class="i">$RefereceSequenceWithNoGaps</span> = <span class="i">RemoveSequenceGaps</span><span class="s">(</span><span class="i">$RefereceSequence</span><span class="s">)</span><span class="sc">;</span>
|
|
461 448 <span class="i">$RefereceSequenceWithNoGapsLen</span> = <span class="k">length</span> <span class="i">$RefereceSequenceWithNoGaps</span><span class="sc">;</span>
|
|
462 449
|
|
463 450 <span class="i">@SpecifiedRegions</span> = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
464 451 <span class="k">push</span> <span class="i">@SpecifiedRegions</span><span class="cm">,</span> <span class="i">@</span>{<span class="i">$OptionsInfo</span>{<span class="w">SpecifiedRegions</span>}}<span class="sc">;</span>
|
|
465 452 <span class="k">if</span> <span class="s">(</span><span class="i">@SpecifiedRegions</span><span class="s">)</span> <span class="s">{</span>
|
|
466 453 <span class="c"># Make sure specified regions are valid...</span>
|
|
467 454 <span class="k">my</span><span class="s">(</span><span class="i">$StartResNum</span><span class="cm">,</span> <span class="i">$EndResNum</span><span class="cm">,</span> <span class="i">$RegionIndex</span><span class="cm">,</span> <span class="i">$RegionNum</span><span class="s">)</span><span class="sc">;</span>
|
|
468 455 <span class="i">$RegionNum</span> = <span class="n">0</span><span class="sc">;</span>
|
|
469 456 <span class="k">for</span> <span class="s">(</span><span class="i">$RegionIndex</span> = <span class="n">0</span><span class="sc">;</span> <span class="i">$RegionIndex</span> <= <span class="i">$#SpecifiedRegions</span><span class="sc">;</span> <span class="i">$RegionIndex</span> += <span class="n">2</span><span class="s">)</span> <span class="s">{</span>
|
|
470 457 <span class="i">$StartResNum</span> = <span class="i">$SpecifiedRegions</span>[<span class="i">$RegionIndex</span>]<span class="sc">;</span>
|
|
471 458 <span class="i">$EndResNum</span> = <span class="i">$SpecifiedRegions</span>[<span class="i">$RegionIndex</span> + <span class="n">1</span>]<span class="sc">;</span>
|
|
472 459 <span class="i">$RegionNum</span>++<span class="sc">;</span>
|
|
473 460 <span class="k">if</span> <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">IgnoreGaps</span>}<span class="s">)</span> <span class="s">{</span>
|
|
474 461 <span class="k">if</span> <span class="s">(</span><span class="i">$StartResNum</span> > <span class="i">$RefereceSequenceWithNoGapsLen</span><span class="s">)</span> <span class="s">{</span>
|
|
475 462 <span class="k">warn</span> <span class="q">"Warning: Ignoring file $SequenceFile: The value specified, $Options{region}, for option \"--region\" is not valid: The start residue number, $StartResNum, must be smaller the sequence length, $RefereceSequenceWithNoGapsLen, of reference sequence ID, $RefereceSequenceID, in region $RegionNum. The reference sequence residue numbers correspond to the sequence with no gaps. Specify \"No\" value for \"-i, --ignoregaps\" option to use residue numbers corresponding to reference sequence including gaps.\n"</span><span class="sc">;</span>
|
|
476 463 <span class="k">next</span> <span class="j">FILELIST</span><span class="sc">;</span>
|
|
477 464 <span class="s">}</span>
|
|
478 465 <span class="k">if</span> <span class="s">(</span><span class="i">$EndResNum</span> > <span class="i">$RefereceSequenceWithNoGapsLen</span><span class="s">)</span> <span class="s">{</span>
|
|
479 466 <span class="k">warn</span> <span class="q">"Warning: Ignoring file $SequenceFile: The value specified, $Options{region}, for option \"--region\" is not valid: The end residue number, $EndResNum, must be smaller the sequence length, $RefereceSequenceWithNoGapsLen, of reference sequence ID, $RefereceSequenceID, in region $RegionNum. The reference sequence residue numbers correspond to the sequence with no gaps. Specify \"No\" value for \"-i, --ignoregaps\" option to use residue numbers corresponding to reference sequence including gaps.\n"</span><span class="sc">;</span>
|
|
480 467 <span class="k">next</span> <span class="j">FILELIST</span><span class="sc">;</span>
|
|
481 468 <span class="s">}</span>
|
|
482 469 <span class="s">}</span>
|
|
483 470 <span class="k">else</span> <span class="s">{</span>
|
|
484 471 <span class="k">if</span> <span class="s">(</span><span class="i">$StartResNum</span> > <span class="i">$RefereceSequenceLen</span><span class="s">)</span> <span class="s">{</span>
|
|
485 472 <span class="k">warn</span> <span class="q">"Warning: Ignoring file $SequenceFile: The value specified, $Options{region}, for option \"--region\" is not valid: The start residue number, $StartResNum, must be smaller the sequence length, $RefereceSequenceLen, of reference sequence ID, $RefereceSequenceID, in region $RegionNum.\n"</span><span class="sc">;</span>
|
|
486 473 <span class="k">next</span> <span class="j">FILELIST</span><span class="sc">;</span>
|
|
487 474 <span class="s">}</span>
|
|
488 475 <span class="k">if</span> <span class="s">(</span><span class="i">$EndResNum</span> > <span class="i">$RefereceSequenceLen</span><span class="s">)</span> <span class="s">{</span>
|
|
489 476 <span class="k">warn</span> <span class="q">"Warning: Ignoring file $SequenceFile: The value specified, $Options{region}, for option \"--region\" is not valid: The end residue number, $EndResNum, must be smaller the sequence length, $RefereceSequenceLen, of reference sequence ID, $RefereceSequenceID, in region $RegionNum.\n"</span><span class="sc">;</span>
|
|
490 477 <span class="k">next</span> <span class="j">FILELIST</span><span class="sc">;</span>
|
|
491 478 <span class="s">}</span>
|
|
492 479 <span class="s">}</span>
|
|
493 480 <span class="k">push</span> <span class="i">@RefereceSequenceRegions</span><span class="cm">,</span> <span class="s">(</span><span class="i">$StartResNum</span><span class="cm">,</span> <span class="i">$EndResNum</span><span class="s">)</span><span class="sc">;</span>
|
|
494 481 <span class="s">}</span>
|
|
495 482 <span class="i">$RefereceSequenceRegionCount</span> = <span class="i">$RegionNum</span><span class="sc">;</span>
|
|
496 483 <span class="s">}</span>
|
|
497 484 <span class="k">else</span> <span class="s">{</span>
|
|
498 485 <span class="c"># Use complete sequence corresponding to reference sequence ID...</span>
|
|
499 486 <span class="k">if</span> <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">IgnoreGaps</span>}<span class="s">)</span> <span class="s">{</span>
|
|
500 487 <span class="k">push</span> <span class="i">@RefereceSequenceRegions</span><span class="cm">,</span> <span class="s">(</span><span class="n">1</span><span class="cm">,</span> <span class="i">$RefereceSequenceWithNoGapsLen</span><span class="s">)</span><span class="sc">;</span>
|
|
501 488 <span class="s">}</span>
|
|
502 489 <span class="k">else</span> <span class="s">{</span>
|
|
503 490 <span class="k">push</span> <span class="i">@RefereceSequenceRegions</span><span class="cm">,</span> <span class="s">(</span><span class="n">1</span><span class="cm">,</span> <span class="i">$RefereceSequenceLen</span><span class="s">)</span><span class="sc">;</span>
|
|
504 491 <span class="s">}</span>
|
|
505 492 <span class="i">$RefereceSequenceRegionCount</span> = <span class="n">1</span><span class="sc">;</span>
|
|
506 493 <span class="s">}</span>
|
|
507 494 <span class="c"># Setup output file names...</span>
|
|
508 495 <span class="i">$FileDir</span> = <span class="q">""</span><span class="sc">;</span> <span class="i">$FileName</span> = <span class="q">""</span><span class="sc">;</span> <span class="i">$FileExt</span> = <span class="q">""</span><span class="sc">;</span>
|
|
509 496 <span class="s">(</span><span class="i">$FileDir</span><span class="cm">,</span> <span class="i">$FileName</span><span class="cm">,</span> <span class="i">$FileExt</span><span class="s">)</span> = <span class="i">ParseFileName</span><span class="s">(</span><span class="i">$SequenceFile</span><span class="s">)</span><span class="sc">;</span>
|
|
510 497 <span class="i">$FileExt</span> = <span class="q">"csv"</span><span class="sc">;</span>
|
|
511 498 <span class="k">if</span> <span class="s">(</span><span class="i">$Options</span>{<span class="w">outdelim</span>} =~ <span class="q">/^tab$/i</span><span class="s">)</span> <span class="s">{</span>
|
|
512 499 <span class="i">$FileExt</span> = <span class="q">"tsv"</span><span class="sc">;</span>
|
|
513 500 <span class="s">}</span>
|
|
514 501 <span class="i">$OutFileExt</span> = <span class="i">$FileExt</span><span class="sc">;</span>
|
|
515 502 <span class="k">if</span> <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">OutFileRoot</span>} && <span class="s">(</span><span class="i">@SequenceFilesList</span> == <span class="n">1</span><span class="s">)</span><span class="s">)</span> <span class="s">{</span>
|
|
516 503 <span class="k">my</span> <span class="s">(</span><span class="i">$RootFileDir</span><span class="cm">,</span> <span class="i">$RootFileName</span><span class="cm">,</span> <span class="i">$RootFileExt</span><span class="s">)</span> = <span class="i">ParseFileName</span><span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">OutFileRoot</span>}<span class="s">)</span><span class="sc">;</span>
|
|
517 504 <span class="k">if</span> <span class="s">(</span><span class="i">$RootFileName</span> && <span class="i">$RootFileExt</span><span class="s">)</span> <span class="s">{</span>
|
|
518 505 <span class="i">$FileName</span> = <span class="i">$RootFileName</span><span class="sc">;</span>
|
|
519 506 <span class="s">}</span>
|
|
520 507 <span class="k">else</span> <span class="s">{</span>
|
|
521 508 <span class="i">$FileName</span> = <span class="i">$OptionsInfo</span>{<span class="w">OutFileRoot</span>}<span class="sc">;</span>
|
|
522 509 <span class="s">}</span>
|
|
523 510 <span class="i">$OutFileRoot</span> = <span class="i">$FileName</span><span class="sc">;</span>
|
|
524 511 <span class="s">}</span>
|
|
525 512 <span class="k">else</span> <span class="s">{</span>
|
|
526 513 <span class="i">$OutFileRoot</span> = <span class="i">$FileName</span><span class="sc">;</span>
|
|
527 514 <span class="s">}</span>
|
|
528 515 <span class="k">if</span> <span class="s">(</span>!<span class="i">$OptionsInfo</span>{<span class="w">OverwriteFiles</span>}<span class="s">)</span> <span class="s">{</span>
|
|
529 516 <span class="k">if</span> <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">CalculatePercentIdentityMatrix</span>}<span class="s">)</span> <span class="s">{</span>
|
|
530 517 <span class="k">if</span> <span class="s">(</span><span class="k">-e</span> <span class="q">"${OutFileRoot}PercentIdentityMatrix.${OutFileExt}"</span><span class="s">)</span> <span class="s">{</span>
|
|
531 518 <span class="k">warn</span> <span class="q">"Warning: Ignoring file $SequenceFile: The file ${OutFileRoot}PercentIdentityMatrix.${OutFileExt} already exists\n"</span><span class="sc">;</span>
|
|
532 519 <span class="k">next</span> <span class="j">FILELIST</span><span class="sc">;</span>
|
|
533 520 <span class="s">}</span>
|
|
534 521 <span class="s">}</span>
|
|
535 522 <span class="k">if</span> <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">PerformResidueFrequencyAnalysis</span>}<span class="s">)</span> <span class="s">{</span>
|
|
536 523 <span class="k">my</span><span class="s">(</span><span class="i">$RegionNum</span><span class="s">)</span><span class="sc">;</span>
|
|
537 524 <span class="k">for</span> <span class="i">$RegionNum</span> <span class="s">(</span><span class="n">1</span> .. <span class="i">$RefereceSequenceRegionCount</span><span class="s">)</span> <span class="s">{</span>
|
|
538 525 <span class="k">if</span> <span class="s">(</span><span class="k">-e</span> <span class="q">"${OutFileRoot}ResidueFrequencyAnalysisRegion${RegionNum}.${OutFileExt}"</span><span class="s">)</span> <span class="s">{</span>
|
|
539 526 <span class="k">warn</span> <span class="q">"Warning: Ignoring file $SequenceFile: The file ${OutFileRoot}ResidueFrequencyAnalysisRegion${RegionNum}.${OutFileExt} already exists\n"</span><span class="sc">;</span>
|
|
540 527 <span class="k">next</span> <span class="j">FILELIST</span><span class="sc">;</span>
|
|
541 528 <span class="s">}</span>
|
|
542 529 <span class="k">if</span> <span class="s">(</span><span class="k">-e</span> <span class="q">"${OutFileRoot}PercentResidueFrequencyAnalysisRegion${RegionNum}.${OutFileExt}"</span><span class="s">)</span> <span class="s">{</span>
|
|
543 530 <span class="k">warn</span> <span class="q">"Warning: Ignoring file $SequenceFile: The file ${OutFileRoot}PercentResidueFrequencyAnalysisRegion${RegionNum}.${OutFileExt} already exists\n"</span><span class="sc">;</span>
|
|
544 531 <span class="k">next</span> <span class="j">FILELIST</span><span class="sc">;</span>
|
|
545 532 <span class="s">}</span>
|
|
546 533 <span class="s">}</span>
|
|
547 534 <span class="s">}</span>
|
|
548 535 <span class="s">}</span>
|
|
549 536
|
|
550 537 <span class="i">$SequenceFilesInfo</span>{<span class="w">FilesOkay</span>}[<span class="i">$Index</span>] = <span class="n">1</span><span class="sc">;</span>
|
|
551 538 <span class="i">$SequenceFilesInfo</span>{<span class="w">OutFileRoot</span>}[<span class="i">$Index</span>] = <span class="i">$OutFileRoot</span><span class="sc">;</span>
|
|
552 539 <span class="i">$SequenceFilesInfo</span>{<span class="w">OutFileExt</span>}[<span class="i">$Index</span>] = <span class="i">$OutFileExt</span><span class="sc">;</span>
|
|
553 540
|
|
554 541 <span class="i">$SequenceFilesInfo</span>{<span class="w">Format</span>}[<span class="i">$Index</span>] = <span class="i">$FileFormat</span><span class="sc">;</span>
|
|
555 542 <span class="i">$SequenceFilesInfo</span>{<span class="w">SequenceCount</span>}[<span class="i">$Index</span>] = <span class="i">$SequenceCount</span><span class="sc">;</span>
|
|
556 543 <span class="i">$SequenceFilesInfo</span>{<span class="w">RefereceSequenceID</span>}[<span class="i">$Index</span>] = <span class="i">$RefereceSequenceID</span><span class="sc">;</span>
|
|
557 544 <span class="i">$SequenceFilesInfo</span>{<span class="w">RefereceSequence</span>}[<span class="i">$Index</span>] = <span class="i">$RefereceSequence</span><span class="sc">;</span>
|
|
558 545 <span class="i">$SequenceFilesInfo</span>{<span class="w">RefereceSequenceLen</span>}[<span class="i">$Index</span>] = <span class="i">$RefereceSequenceLen</span><span class="sc">;</span>
|
|
559 546 <span class="i">$SequenceFilesInfo</span>{<span class="w">RefereceSequenceWithNoGaps</span>}[<span class="i">$Index</span>] = <span class="i">$RefereceSequenceWithNoGaps</span><span class="sc">;</span>
|
|
560 547 <span class="i">$SequenceFilesInfo</span>{<span class="w">RefereceSequenceWithNoGapsLen</span>}[<span class="i">$Index</span>] = <span class="i">$RefereceSequenceWithNoGapsLen</span><span class="sc">;</span>
|
|
561 548 <span class="k">push</span> <span class="i">@</span>{<span class="i">$SequenceFilesInfo</span>{<span class="w">RefereceSequenceRegions</span>}[<span class="i">$Index</span>]}<span class="cm">,</span> <span class="i">@RefereceSequenceRegions</span><span class="sc">;</span>
|
|
562 549 <span class="i">$SequenceFilesInfo</span>{<span class="w">RefereceSequenceRegionCount</span>}[<span class="i">$Index</span>] = <span class="i">$RefereceSequenceRegionCount</span><span class="sc">;</span>
|
|
563 550
|
|
564 551 <span class="c"># Setup residue codes...</span>
|
|
565 552 <span class="i">SetupSequenceFileResidueCodes</span><span class="s">(</span><span class="i">$SequenceDataRef</span><span class="cm">,</span> <span class="i">$Index</span><span class="s">)</span><span class="sc">;</span>
|
|
566 553 <span class="s">}</span>
|
|
567 554 <span class="s">}</span>
|
|
568 555
|
|
569 <a name="SetupSequenceFileResidueCodes-"></a> 556 <span class="k">sub </span><span class="m">SetupSequenceFileResidueCodes</span> <span class="s">{</span>
|
|
570 557 <span class="k">my</span><span class="s">(</span><span class="i">$SequenceDataRef</span><span class="cm">,</span> <span class="i">$FileIndex</span><span class="s">)</span> = <span class="i">@_</span><span class="sc">;</span>
|
|
571 558 <span class="k">my</span><span class="s">(</span><span class="i">$Residue</span><span class="cm">,</span> <span class="i">@ResidueCodesList</span><span class="cm">,</span> <span class="i">%ResidueCodesMap</span><span class="s">)</span><span class="sc">;</span>
|
|
572 559
|
|
573 560 <span class="c"># Initialize</span>
|
|
574 561 <span class="i">@</span>{<span class="i">$SequenceFilesInfo</span>{<span class="w">ResidueCodes</span>}[<span class="i">$FileIndex</span>]} = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
575 562
|
|
576 563 <span class="i">%ResidueCodesMap</span> = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
577 564 <span class="i">@ResidueCodesList</span> = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
578 565
|
|
579 566 <span class="c"># Setup standard amino acids and nucleic acids one letter codes...</span>
|
|
580 567 <span class="k">if</span> <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">RegionResiduesMode</span>} =~ <span class="q">/^AminoAcids$/i</span><span class="s">)</span> <span class="s">{</span>
|
|
581 568 <span class="i">@ResidueCodesList</span> = <span class="i">AminoAcids::GetAminoAcids</span><span class="s">(</span><span class="q">'OneLetterCode'</span><span class="s">)</span><span class="sc">;</span>
|
|
582 569 <span class="s">}</span>
|
|
583 570 <span class="k">elsif</span> <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">RegionResiduesMode</span>} =~ <span class="q">/^NucleicAcids$/i</span><span class="s">)</span> <span class="s">{</span>
|
|
584 571 <span class="k">push</span> <span class="i">@ResidueCodesList</span><span class="cm">,</span> <span class="s">(</span><span class="q">'A'</span><span class="cm">,</span> <span class="q">'G'</span><span class="cm">,</span> <span class="q">'T'</span><span class="cm">,</span> <span class="q">'U'</span><span class="cm">,</span> <span class="q">'C'</span><span class="s">)</span><span class="sc">;</span>
|
|
585 572 <span class="s">}</span>
|
|
586 573 <span class="k">push</span> <span class="i">@ResidueCodesList</span><span class="cm">,</span> <span class="q">'Gap'</span><span class="sc">;</span>
|
|
587 574 <span class="k">for</span> <span class="i">$Residue</span> <span class="s">(</span><span class="i">@ResidueCodesList</span><span class="s">)</span> <span class="s">{</span>
|
|
588 575 <span class="i">$ResidueCodesMap</span>{<span class="i">$Residue</span>} = <span class="i">$Residue</span><span class="sc">;</span>
|
|
589 576 <span class="s">}</span>
|
|
590 577
|
|
591 578 <span class="c"># Go over all the residues in all the sequences and add missing ones to the list...</span>
|
|
592 579 <span class="k">my</span><span class="s">(</span><span class="i">$ID</span><span class="cm">,</span> <span class="i">$Sequence</span><span class="cm">,</span> <span class="i">$ResIndex</span><span class="s">)</span><span class="sc">;</span>
|
|
593 580 <span class="k">for</span> <span class="i">$ID</span> <span class="s">(</span><span class="i">@</span>{<span class="i">$SequenceDataRef</span>->{<span class="w">IDs</span>}}<span class="s">)</span> <span class="s">{</span>
|
|
594 581 <span class="i">$Sequence</span> = <span class="i">$SequenceDataRef</span>->{<span class="w">Sequence</span>}{<span class="i">$ID</span>}<span class="sc">;</span>
|
|
595 582 <span class="j">RES:</span> <span class="k">for</span> <span class="i">$ResIndex</span> <span class="s">(</span><span class="n">0</span> .. <span class="s">(</span><span class="k">length</span><span class="s">(</span><span class="i">$Sequence</span><span class="s">)</span> - <span class="n">1</span><span class="s">)</span><span class="s">)</span> <span class="s">{</span>
|
|
596 583 <span class="i">$Residue</span> = <span class="k">substr</span><span class="s">(</span><span class="i">$Sequence</span><span class="cm">,</span> <span class="i">$ResIndex</span><span class="cm">,</span> <span class="n">1</span><span class="s">)</span><span class="sc">;</span>
|
|
597 584 <span class="k">if</span> <span class="s">(</span><span class="i">IsGapResidue</span><span class="s">(</span><span class="i">$Residue</span><span class="s">)</span><span class="s">)</span> <span class="s">{</span>
|
|
598 585 <span class="k">next</span> <span class="j">RES</span><span class="sc">;</span>
|
|
599 586 <span class="s">}</span>
|
|
600 587 <span class="k">if</span> <span class="s">(</span><span class="k">exists</span> <span class="i">$ResidueCodesMap</span>{<span class="i">$Residue</span>}<span class="s">)</span> <span class="s">{</span>
|
|
601 588 <span class="k">next</span> <span class="j">RES</span><span class="sc">;</span>
|
|
602 589 <span class="s">}</span>
|
|
603 590 <span class="k">push</span> <span class="i">@ResidueCodesList</span><span class="cm">,</span> <span class="i">$Residue</span><span class="sc">;</span>
|
|
604 591 <span class="i">$ResidueCodesMap</span>{<span class="i">$Residue</span>} = <span class="i">$Residue</span><span class="sc">;</span>
|
|
605 592 <span class="s">}</span>
|
|
606 593 <span class="s">}</span>
|
|
607 594 <span class="k">push</span> <span class="i">@</span>{<span class="i">$SequenceFilesInfo</span>{<span class="w">ResidueCodes</span>}[<span class="i">$FileIndex</span>]}<span class="cm">,</span> <span class="i">@ResidueCodesList</span><span class="sc">;</span>
|
|
608 595 <span class="s">}</span>
|
|
609 596
|
|
610 597 <span class="c"># Setup regions data for performing residue frequency analysis...</span>
|
|
611 <a name="SetupSequenceRegionsData-"></a> 598 <span class="k">sub </span><span class="m">SetupSequenceRegionsData</span> <span class="s">{</span>
|
|
612 599 <span class="k">my</span><span class="s">(</span><span class="i">$Index</span><span class="cm">,</span> <span class="i">$RefereceSequence</span><span class="cm">,</span> <span class="i">$RefereceSequenceLen</span><span class="cm">,</span> <span class="i">$RegionID</span><span class="cm">,</span> <span class="i">$StartResNum</span><span class="cm">,</span> <span class="i">$EndResNum</span><span class="cm">,</span> <span class="i">$RegionIndex</span><span class="cm">,</span> <span class="i">$RegionNum</span><span class="cm">,</span> <span class="i">$NoGapResNum</span><span class="cm">,</span> <span class="i">$ResNum</span><span class="cm">,</span> <span class="i">$ResIndex</span><span class="cm">,</span> <span class="i">$Residue</span><span class="cm">,</span> <span class="i">$ResidueCode</span><span class="cm">,</span> <span class="i">@RefereceSequenceRegions</span><span class="s">)</span><span class="sc">;</span>
|
|
613 600
|
|
614 601
|
|
615 602 <span class="i">@</span>{<span class="i">$SequenceFilesInfo</span>{<span class="w">RefereceSequenceResNums</span>}} = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
616 603 <span class="i">@</span>{<span class="i">$SequenceFilesInfo</span>{<span class="w">RegionsData</span>}} = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
617 604
|
|
618 605 <span class="j">FILELIST:</span> <span class="k">for</span> <span class="i">$Index</span> <span class="s">(</span><span class="n">0</span> .. <span class="i">$#SequenceFilesList</span><span class="s">)</span> <span class="s">{</span>
|
|
619 606 <span class="i">%</span>{<span class="i">$SequenceFilesInfo</span>{<span class="w">RefereceSequenceResNums</span>}[<span class="i">$Index</span>]{<span class="w">IsGap</span>}} = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
620 607 <span class="i">%</span>{<span class="i">$SequenceFilesInfo</span>{<span class="w">RefereceSequenceResNums</span>}[<span class="i">$Index</span>]{<span class="w">NoGapToGap</span>}} = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
621 608 <span class="i">%</span>{<span class="i">$SequenceFilesInfo</span>{<span class="w">RefereceSequenceResNums</span>}[<span class="i">$Index</span>]{<span class="w">GapToNoGap</span>}} = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
622 609 <span class="i">%</span>{<span class="i">$SequenceFilesInfo</span>{<span class="w">RegionsData</span>}[<span class="i">$Index</span>]} = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
623 610 <span class="i">@</span>{<span class="i">$SequenceFilesInfo</span>{<span class="w">RegionsData</span>}[<span class="i">$Index</span>]{<span class="w">RegionIDs</span>}} = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
624 611
|
|
625 612 <span class="k">if</span> <span class="s">(</span>!<span class="i">$SequenceFilesInfo</span>{<span class="w">FilesOkay</span>}[<span class="i">$Index</span>]<span class="s">)</span> <span class="s">{</span>
|
|
626 613 <span class="k">next</span> <span class="j">FILELIST</span><span class="sc">;</span>
|
|
627 614 <span class="s">}</span>
|
|
628 615 <span class="k">if</span> <span class="s">(</span>!<span class="i">$OptionsInfo</span>{<span class="w">PerformResidueFrequencyAnalysis</span>}<span class="s">)</span> <span class="s">{</span>
|
|
629 616 <span class="k">next</span> <span class="j">FILELIST</span><span class="sc">;</span>
|
|
630 617 <span class="s">}</span>
|
|
631 618
|
|
632 619 <span class="i">$RefereceSequence</span> = <span class="i">$SequenceFilesInfo</span>{<span class="w">RefereceSequence</span>}[<span class="i">$Index</span>]<span class="sc">;</span>
|
|
633 620 <span class="i">$RefereceSequenceLen</span> = <span class="i">$SequenceFilesInfo</span>{<span class="w">RefereceSequenceLen</span>}[<span class="i">$Index</span>]<span class="sc">;</span>
|
|
634 621
|
|
635 622 <span class="c"># Setup residue number mapping and gap status for refernece sequence...</span>
|
|
636 623 <span class="i">$NoGapResNum</span> = <span class="n">0</span><span class="sc">;</span>
|
|
637 624 <span class="i">$ResNum</span> = <span class="n">0</span><span class="sc">;</span>
|
|
638 625 <span class="k">for</span> <span class="i">$ResIndex</span> <span class="s">(</span><span class="n">0</span> .. <span class="s">(</span><span class="i">$RefereceSequenceLen</span> - <span class="n">1</span><span class="s">)</span><span class="s">)</span> <span class="s">{</span>
|
|
639 626 <span class="i">$ResNum</span>++<span class="sc">;</span>
|
|
640 627 <span class="i">$Residue</span> = <span class="k">substr</span><span class="s">(</span><span class="i">$RefereceSequence</span><span class="cm">,</span> <span class="i">$ResIndex</span><span class="cm">,</span> <span class="n">1</span><span class="s">)</span><span class="sc">;</span>
|
|
641 628 <span class="i">$SequenceFilesInfo</span>{<span class="w">RefereceSequenceResNums</span>}[<span class="i">$Index</span>]{<span class="w">IsGap</span>}{<span class="i">$ResNum</span>} = <span class="n">1</span><span class="sc">;</span>
|
|
642 629 <span class="k">if</span> <span class="s">(</span>!<span class="i">IsGapResidue</span><span class="s">(</span><span class="i">$Residue</span><span class="s">)</span><span class="s">)</span> <span class="s">{</span>
|
|
643 630 <span class="i">$NoGapResNum</span>++<span class="sc">;</span>
|
|
644 631 <span class="i">$SequenceFilesInfo</span>{<span class="w">RefereceSequenceResNums</span>}[<span class="i">$Index</span>]{<span class="w">IsGap</span>}{<span class="i">$ResNum</span>} = <span class="n">0</span><span class="sc">;</span>
|
|
645 632 <span class="i">$SequenceFilesInfo</span>{<span class="w">RefereceSequenceResNums</span>}[<span class="i">$Index</span>]{<span class="w">NoGapToGap</span>}{<span class="i">$NoGapResNum</span>} = <span class="i">$ResNum</span><span class="sc">;</span>
|
|
646 633 <span class="i">$SequenceFilesInfo</span>{<span class="w">RefereceSequenceResNums</span>}[<span class="i">$Index</span>]{<span class="w">GapToNoGap</span>}{<span class="i">$ResNum</span>} = <span class="i">$NoGapResNum</span><span class="sc">;</span>
|
|
647 634 <span class="s">}</span>
|
|
648 635 <span class="s">}</span>
|
|
649 636 <span class="c"># Map residue numbers for specified regions to the reference residue in input sequence/alignment files</span>
|
|
650 637 <span class="i">$RegionNum</span> = <span class="n">0</span><span class="sc">;</span>
|
|
651 638 <span class="i">@RefereceSequenceRegions</span> = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
652 639 <span class="k">push</span> <span class="i">@RefereceSequenceRegions</span><span class="cm">,</span> <span class="i">@</span>{<span class="i">$SequenceFilesInfo</span>{<span class="w">RefereceSequenceRegions</span>}[<span class="i">$Index</span>]}<span class="sc">;</span>
|
|
653 640 <span class="k">for</span> <span class="s">(</span><span class="i">$RegionIndex</span> = <span class="n">0</span><span class="sc">;</span> <span class="i">$RegionIndex</span> <= <span class="i">$#RefereceSequenceRegions</span><span class="sc">;</span> <span class="i">$RegionIndex</span> += <span class="n">2</span><span class="s">)</span> <span class="s">{</span>
|
|
654 641 <span class="i">$StartResNum</span> = <span class="i">$RefereceSequenceRegions</span>[<span class="i">$RegionIndex</span>]<span class="sc">;</span>
|
|
655 642 <span class="i">$EndResNum</span> = <span class="i">$RefereceSequenceRegions</span>[<span class="i">$RegionIndex</span> + <span class="n">1</span>]<span class="sc">;</span>
|
|
656 643 <span class="i">$RegionNum</span>++<span class="sc">;</span>
|
|
657 644 <span class="i">$RegionID</span> = <span class="q">"Region${RegionNum}"</span><span class="sc">;</span>
|
|
658 645 <span class="k">if</span> <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">IgnoreGaps</span>}<span class="s">)</span> <span class="s">{</span>
|
|
659 646 <span class="c"># Map residue numbers to the reference sequence residue numbers to account for any ignored gaps...</span>
|
|
660 647 <span class="i">$StartResNum</span> = <span class="i">$SequenceFilesInfo</span>{<span class="w">RefereceSequenceResNums</span>}[<span class="i">$Index</span>]{<span class="w">NoGapToGap</span>}{<span class="i">$StartResNum</span>}<span class="sc">;</span>
|
|
661 648 <span class="i">$EndResNum</span> = <span class="i">$SequenceFilesInfo</span>{<span class="w">RefereceSequenceResNums</span>}[<span class="i">$Index</span>]{<span class="w">NoGapToGap</span>}{<span class="i">$EndResNum</span>}<span class="sc">;</span>
|
|
662 649 <span class="s">}</span>
|
|
663 650 <span class="k">push</span> <span class="i">@</span>{<span class="i">$SequenceFilesInfo</span>{<span class="w">RegionsData</span>}[<span class="i">$Index</span>]{<span class="w">RegionIDs</span>}}<span class="cm">,</span> <span class="i">$RegionID</span><span class="sc">;</span>
|
|
664 651 <span class="i">$SequenceFilesInfo</span>{<span class="w">RegionsData</span>}[<span class="i">$Index</span>]{<span class="i">$RegionID</span>}{<span class="w">StartResNum</span>} = <span class="i">$StartResNum</span><span class="sc">;</span>
|
|
665 652 <span class="i">$SequenceFilesInfo</span>{<span class="w">RegionsData</span>}[<span class="i">$Index</span>]{<span class="i">$RegionID</span>}{<span class="w">EndResNum</span>} = <span class="i">$EndResNum</span><span class="sc">;</span>
|
|
666 653
|
|
667 654 <span class="c"># Initialize data for residue codes...</span>
|
|
668 655 <span class="k">for</span> <span class="i">$ResNum</span> <span class="s">(</span><span class="i">$StartResNum</span> .. <span class="i">$EndResNum</span><span class="s">)</span> <span class="s">{</span>
|
|
669 656 <span class="k">for</span> <span class="i">$ResidueCode</span> <span class="s">(</span><span class="i">@</span>{<span class="i">$SequenceFilesInfo</span>{<span class="w">ResidueCodes</span>}[<span class="i">$Index</span>]}<span class="s">)</span> <span class="s">{</span>
|
|
670 657 <span class="i">$SequenceFilesInfo</span>{<span class="w">RegionsData</span>}[<span class="i">$Index</span>]{<span class="i">$RegionID</span>}{<span class="w">Count</span>}{<span class="i">$ResNum</span>}{<span class="i">$ResidueCode</span>} = <span class="n">0</span><span class="sc">;</span>
|
|
671 658 <span class="s">}</span>
|
|
672 659 <span class="s">}</span>
|
|
673 660 <span class="s">}</span>
|
|
674 661 <span class="s">}</span>
|
|
675 662 <span class="s">}</span>
|
|
676 663
|
|
677 664 <span class="c"># Setup script usage and retrieve command line arguments specified using various options...</span>
|
|
678 <a name="SetupScriptUsage-"></a> 665 <span class="k">sub </span><span class="m">SetupScriptUsage</span> <span class="s">{</span>
|
|
679 666
|
|
680 667 <span class="c"># Retrieve all the options...</span>
|
|
681 668 <span class="i">%Options</span> = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
682 669 <span class="i">$Options</span>{<span class="w">ignoregaps</span>} = <span class="q">'yes'</span><span class="sc">;</span>
|
|
683 670 <span class="i">$Options</span>{<span class="w">regionresiduesmode</span>} = <span class="q">'None'</span><span class="sc">;</span>
|
|
684 671 <span class="i">$Options</span>{<span class="w">mode</span>} = <span class="q">'PercentIdentityMatrix'</span><span class="sc">;</span>
|
|
685 672 <span class="i">$Options</span>{<span class="w">outdelim</span>} = <span class="q">'comma'</span><span class="sc">;</span>
|
|
686 673 <span class="i">$Options</span>{<span class="w">precision</span>} = <span class="n">2</span><span class="sc">;</span>
|
|
687 674 <span class="i">$Options</span>{<span class="w">quote</span>} = <span class="q">'yes'</span><span class="sc">;</span>
|
|
688 675 <span class="i">$Options</span>{<span class="w">referencesequence</span>} = <span class="q">'UseFirstSequenceID'</span><span class="sc">;</span>
|
|
689 676 <span class="i">$Options</span>{<span class="w">region</span>} = <span class="q">'UseCompleteSequence'</span><span class="sc">;</span>
|
|
690 677
|
|
691 678 <span class="k">if</span> <span class="s">(</span>!<span class="i">GetOptions</span><span class="s">(</span>\<span class="i">%Options</span><span class="cm">,</span> <span class="q">"help|h"</span><span class="cm">,</span> <span class="q">"ignoregaps|i=s"</span><span class="cm">,</span> <span class="q">"mode|m=s"</span><span class="cm">,</span> <span class="q">"outdelim=s"</span><span class="cm">,</span> <span class="q">"overwrite|o"</span><span class="cm">,</span> <span class="q">"precision|p=i"</span><span class="cm">,</span> <span class="q">"quote|q=s"</span><span class="cm">,</span> <span class="q">"referencesequence=s"</span><span class="cm">,</span> <span class="q">"region=s"</span><span class="cm">,</span> <span class="q">"regionresiduesmode=s"</span><span class="cm">,</span> <span class="q">"root|r=s"</span><span class="cm">,</span> <span class="q">"workingdir|w=s"</span><span class="s">)</span><span class="s">)</span> <span class="s">{</span>
|
|
692 679 <span class="k">die</span> <span class="q">"\nTo get a list of valid options and their values, use \"$ScriptName -h\" or\n\"perl -S $ScriptName -h\" command and try again...\n"</span><span class="sc">;</span>
|
|
693 680 <span class="s">}</span>
|
|
694 681 <span class="k">if</span> <span class="s">(</span><span class="i">$Options</span>{<span class="w">workingdir</span>}<span class="s">)</span> <span class="s">{</span>
|
|
695 682 <span class="k">if</span> <span class="s">(</span>! <span class="k">-d</span> <span class="i">$Options</span>{<span class="w">workingdir</span>}<span class="s">)</span> <span class="s">{</span>
|
|
696 683 <span class="k">die</span> <span class="q">"Error: The value specified, $Options{workingdir}, for option \"-w --workingdir\" is not a directory name.\n"</span><span class="sc">;</span>
|
|
697 684 <span class="s">}</span>
|
|
698 685 <span class="k">chdir</span> <span class="i">$Options</span>{<span class="w">workingdir</span>} <span class="k">or</span> <span class="k">die</span> <span class="q">"Error: Couldn't chdir $Options{workingdir}: $! \n"</span><span class="sc">;</span>
|
|
699 686 <span class="s">}</span>
|
|
700 687 <span class="k">if</span> <span class="s">(</span><span class="i">$Options</span>{<span class="w">ignoregaps</span>} !~ <span class="q">/^(yes|no)$/i</span><span class="s">)</span> <span class="s">{</span>
|
|
701 688 <span class="k">die</span> <span class="q">"Error: The value specified, $Options{ignoregaps}, for option \"-q --quote\" is not valid. Allowed values: yes or no\n"</span><span class="sc">;</span>
|
|
702 689 <span class="s">}</span>
|
|
703 690 <span class="k">if</span> <span class="s">(</span><span class="i">$Options</span>{<span class="w">regionresiduesmode</span>} !~ <span class="q">/^(AminoAcids|NucleicAcids|None)$/i</span><span class="s">)</span> <span class="s">{</span>
|
|
704 691 <span class="k">die</span> <span class="q">"Error: The value specified, $Options{regionresiduesmode}, for option \"--regionresiduesmode\" is not valid. Allowed values: AminoAcids, NucleicAcids or None\n"</span><span class="sc">;</span>
|
|
705 692 <span class="s">}</span>
|
|
706 693 <span class="k">if</span> <span class="s">(</span><span class="i">$Options</span>{<span class="w">mode</span>} !~ <span class="q">/^(PercentIdentityMatrix|ResidueFrequencyAnalysis|All)$/i</span><span class="s">)</span> <span class="s">{</span>
|
|
707 694 <span class="k">die</span> <span class="q">"Error: The value specified, $Options{mode}, for option \"-m --mode\" is not valid. Allowed values: PercentIdentityMatrix, ResidueFrequencyAnalysis or All\n"</span><span class="sc">;</span>
|
|
708 695 <span class="s">}</span>
|
|
709 696 <span class="k">if</span> <span class="s">(</span><span class="i">$Options</span>{<span class="w">outdelim</span>} !~ <span class="q">/^(comma|semicolon|tab)$/i</span><span class="s">)</span> <span class="s">{</span>
|
|
710 697 <span class="k">die</span> <span class="q">"Error: The value specified, $Options{outdelim}, for option \"--outdelim\" is not valid. Allowed values: comma, tab, or semicolon\n"</span><span class="sc">;</span>
|
|
711 698 <span class="s">}</span>
|
|
712 699 <span class="k">if</span> <span class="s">(</span><span class="i">$Options</span>{<span class="w">quote</span>} !~ <span class="q">/^(yes|no)$/i</span><span class="s">)</span> <span class="s">{</span>
|
|
713 700 <span class="k">die</span> <span class="q">"Error: The value specified, $Options{quote}, for option \"-q --quote\" is not valid. Allowed values: yes or no\n"</span><span class="sc">;</span>
|
|
714 701 <span class="s">}</span>
|
|
715 702 <span class="k">if</span> <span class="s">(</span>!<span class="i">IsPositiveInteger</span><span class="s">(</span><span class="i">$Options</span>{<span class="w">precision</span>}<span class="s">)</span><span class="s">)</span> <span class="s">{</span>
|
|
716 703 <span class="k">die</span> <span class="q">"Error: The value specified, $Options{precision}, for option \"-p --precision\" is not valid. Allowed values: > 0 \n"</span><span class="sc">;</span>
|
|
717 704 <span class="s">}</span>
|
|
718 705 <span class="s">}</span>
|
|
719 706
|
|
720 <a name="EOF-"></a></pre>
|
|
721 <p> </p>
|
|
722 <br />
|
|
723 <center>
|
|
724 <img src="../../../images/h2o2.png">
|
|
725 </center>
|
|
726 </body>
|
|
727 </html>
|