0
|
1 <html>
|
|
2 <head>
|
|
3 <title>MayaChemTools:Code:ExtractFromSequenceFiles.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: ExtractFromSequenceFiles.pl,v $</span>
|
|
17 4 <span class="c"># $Date: 2015/02/28 20:46:19 $</span>
|
|
18 5 <span class="c"># $Revision: 1.23 $</span>
|
|
19 6 <span class="c">#</span>
|
|
20 7 <span class="c"># Author: Manish Sud <msud@san.rr.com></span>
|
|
21 8 <span class="c">#</span>
|
|
22 9 <span class="c"># Copyright (C) 2015 Manish Sud. All rights reserved.</span>
|
|
23 10 <span class="c">#</span>
|
|
24 11 <span class="c"># This file is part of MayaChemTools.</span>
|
|
25 12 <span class="c">#</span>
|
|
26 13 <span class="c"># MayaChemTools is free software; you can redistribute it and/or modify it under</span>
|
|
27 14 <span class="c"># the terms of the GNU Lesser General Public License as published by the Free</span>
|
|
28 15 <span class="c"># Software Foundation; either version 3 of the License, or (at your option) any</span>
|
|
29 16 <span class="c"># later version.</span>
|
|
30 17 <span class="c">#</span>
|
|
31 18 <span class="c"># MayaChemTools is distributed in the hope that it will be useful, but without</span>
|
|
32 19 <span class="c"># any warranty; without even the implied warranty of merchantability of fitness</span>
|
|
33 20 <span class="c"># for a particular purpose. See the GNU Lesser General Public License for more</span>
|
|
34 21 <span class="c"># details.</span>
|
|
35 22 <span class="c">#</span>
|
|
36 23 <span class="c"># You should have received a copy of the GNU Lesser General Public License</span>
|
|
37 24 <span class="c"># along with MayaChemTools; if not, see <http://www.gnu.org/licenses/> or</span>
|
|
38 25 <span class="c"># write to the Free Software Foundation Inc., 59 Temple Place, Suite 330,</span>
|
|
39 26 <span class="c"># Boston, MA, 02111-1307, USA.</span>
|
|
40 27 <span class="c">#</span>
|
|
41 28
|
|
42 29 <span class="k">use</span> <span class="w">strict</span><span class="sc">;</span>
|
|
43 30 <span class="k">use</span> <span class="w">FindBin</span><span class="sc">;</span> <span class="k">use</span> <span class="w">lib</span> <span class="q">"$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
|
|
52 39 <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>
|
|
53 40
|
|
54 41 <span class="c"># Autoflush STDOUT</span>
|
|
55 42 <span class="i">$|</span> = <span class="n">1</span><span class="sc">;</span>
|
|
56 43
|
|
57 44 <span class="c"># Starting message...</span>
|
|
58 45 <span class="i">$ScriptName</span> = <span class="i">basename</span><span class="s">(</span><span class="i">$0</span><span class="s">)</span><span class="sc">;</span>
|
|
59 46 <span class="k">print</span> <span class="q">"\n$ScriptName: Starting...\n\n"</span><span class="sc">;</span>
|
|
60 47 <span class="i">$StartTime</span> = <span class="w">new</span> <span class="w">Benchmark</span><span class="sc">;</span>
|
|
61 48
|
|
62 49 <span class="c"># Setup script usage message...</span>
|
|
63 50 <span class="i">SetupScriptUsage</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
64 51 <span class="k">if</span> <span class="s">(</span><span class="i">$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>
|
|
65 52 <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>
|
|
66 53 <span class="s">}</span>
|
|
67 54
|
|
68 55 <span class="c"># Expand wild card file names...</span>
|
|
69 56 <span class="k">my</span><span class="s">(</span><span class="i">@SequenceFilesList</span><span class="s">)</span><span class="sc">;</span>
|
|
70 57 <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>
|
|
71 58
|
|
72 59 <span class="c"># Process options...</span>
|
|
73 60 <span class="k">print</span> <span class="q">"Processing options...\n"</span><span class="sc">;</span>
|
|
74 61 <span class="k">my</span><span class="s">(</span><span class="i">%OptionsInfo</span><span class="s">)</span><span class="sc">;</span>
|
|
75 62 <span class="i">ProcessOptions</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
76 63
|
|
77 64 <span class="c"># Set up information about input files...</span>
|
|
78 65 <span class="k">print</span> <span class="q">"Checking input sequence file(s)...\n"</span><span class="sc">;</span>
|
|
79 66 <span class="k">my</span><span class="s">(</span><span class="i">%SequenceFilesInfo</span><span class="s">)</span><span class="sc">;</span>
|
|
80 67 <span class="i">RetrieveSequenceFilesInfo</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
81 68
|
|
82 69 <span class="c"># Process input files..</span>
|
|
83 70 <span class="k">my</span><span class="s">(</span><span class="i">$FileIndex</span><span class="s">)</span><span class="sc">;</span>
|
|
84 71 <span class="k">if</span> <span class="s">(</span><span class="i">@SequenceFilesList</span> > <span class="n">1</span><span class="s">)</span> <span class="s">{</span>
|
|
85 72 <span class="k">print</span> <span class="q">"\nProcessing sequence files...\n"</span><span class="sc">;</span>
|
|
86 73 <span class="s">}</span>
|
|
87 74 <span class="k">for</span> <span class="i">$FileIndex</span> <span class="s">(</span><span class="n">0</span> .. <span class="i">$#SequenceFilesList</span><span class="s">)</span> <span class="s">{</span>
|
|
88 75 <span class="k">if</span> <span class="s">(</span><span class="i">$SequenceFilesInfo</span>{<span class="w">FilesOkay</span>}[<span class="i">$FileIndex</span>]<span class="s">)</span> <span class="s">{</span>
|
|
89 76 <span class="k">print</span> <span class="q">"\nProcessing file $SequenceFilesList[$FileIndex]...\n"</span><span class="sc">;</span>
|
|
90 77 <span class="i">ExtractFromSequenceFiles</span><span class="s">(</span><span class="i">$FileIndex</span><span class="s">)</span><span class="sc">;</span>
|
|
91 78 <span class="s">}</span>
|
|
92 79 <span class="s">}</span>
|
|
93 80 <span class="k">print</span> <span class="q">"\n$ScriptName:Done...\n\n"</span><span class="sc">;</span>
|
|
94 81
|
|
95 82 <span class="i">$EndTime</span> = <span class="w">new</span> <span class="w">Benchmark</span><span class="sc">;</span>
|
|
96 83 <span class="i">$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>
|
|
97 84 <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">$TotalTime</span><span class="s">)</span><span class="cm">,</span> <span class="q">"\n"</span><span class="sc">;</span>
|
|
98 85
|
|
99 86 <span class="c">###############################################################################</span>
|
|
100 87
|
|
101 88 <span class="c"># Extract from sequence files...</span>
|
|
102 <a name="ExtractFromSequenceFiles-"></a> 89 <span class="k">sub </span><span class="m">ExtractFromSequenceFiles</span> <span class="s">{</span>
|
|
103 90 <span class="k">my</span><span class="s">(</span><span class="i">$FileIndex</span><span class="s">)</span> = <span class="i">@_</span><span class="sc">;</span>
|
|
104 91 <span class="k">my</span><span class="s">(</span><span class="i">$OutSequenceFile</span><span class="cm">,</span> <span class="i">$SequenceFile</span><span class="cm">,</span> <span class="i">$SequenceDataRef</span><span class="cm">,</span> <span class="i">$SpecifiedSequenceDataRef</span><span class="s">)</span><span class="sc">;</span>
|
|
105 92
|
|
106 93 <span class="c"># Read sequence file...</span>
|
|
107 94 <span class="i">$SequenceFile</span> = <span class="i">$SequenceFilesList</span>[<span class="i">$FileIndex</span>]<span class="sc">;</span>
|
|
108 95 <span class="k">open</span> <span class="w">SEQUENCEFILE</span><span class="cm">,</span> <span class="q">"$SequenceFile"</span> <span class="k">or</span> <span class="k">die</span> <span class="q">"Error: Can't open $SequenceFile: $! \n"</span><span class="sc">;</span>
|
|
109 96 <span class="i">$SequenceDataRef</span> = <span class="i">ReadSequenceFile</span><span class="s">(</span><span class="i">$SequenceFile</span><span class="s">)</span><span class="sc">;</span>
|
|
110 97 <span class="k">close</span> <span class="w">SEQUENCEFILE</span><span class="sc">;</span>
|
|
111 98
|
|
112 99 <span class="i">$OutSequenceFile</span> = <span class="i">$SequenceFilesInfo</span>{<span class="w">OutFile</span>}[<span class="i">$FileIndex</span>]<span class="sc">;</span>
|
|
113 100 <span class="k">print</span> <span class="q">"Generating sequence file $OutSequenceFile...\n"</span><span class="sc">;</span>
|
|
114 101
|
|
115 102 <span class="c"># Retrieve sequence data for specified sequences...</span>
|
|
116 103 <span class="i">$SpecifiedSequenceDataRef</span> = <span class="i">GetSpecifiedSequenceData</span><span class="s">(</span><span class="i">$SequenceDataRef</span><span class="s">)</span><span class="sc">;</span>
|
|
117 104
|
|
118 105 <span class="c"># Handle gaps...</span>
|
|
119 106 <span class="k">if</span> <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">IgnoreGaps</span>}<span class="s">)</span> <span class="s">{</span>
|
|
120 107 <span class="k">if</span> <span class="s">(</span><span class="i">@</span>{<span class="i">$SpecifiedSequenceDataRef</span>->{<span class="w">IDs</span>}} > <span class="n">1</span><span class="s">)</span> <span class="s">{</span>
|
|
121 108 <span class="k">if</span> <span class="s">(</span><span class="i">AreSequenceLengthsIdentical</span><span class="s">(</span><span class="i">$SpecifiedSequenceDataRef</span><span class="s">)</span><span class="s">)</span> <span class="s">{</span>
|
|
122 109 <span class="i">$SpecifiedSequenceDataRef</span> = <span class="i">RemoveSequenceAlignmentGapColumns</span><span class="s">(</span><span class="i">$SpecifiedSequenceDataRef</span><span class="s">)</span><span class="sc">;</span>
|
|
123 110 <span class="s">}</span>
|
|
124 111 <span class="s">}</span>
|
|
125 112 <span class="k">else</span> <span class="s">{</span>
|
|
126 113 <span class="c"># Remove the gaps from the sequence...</span>
|
|
127 114 <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="s">)</span><span class="sc">;</span>
|
|
128 115 <span class="i">$ID</span> = <span class="i">$SpecifiedSequenceDataRef</span>->{<span class="w">IDs</span>}[<span class="n">0</span>]<span class="sc">;</span>
|
|
129 116 <span class="i">$Sequence</span> = <span class="i">$SpecifiedSequenceDataRef</span>->{<span class="w">Sequence</span>}{<span class="i">$ID</span>}<span class="sc">;</span>
|
|
130 117 <span class="i">$SpecifiedSequenceDataRef</span>->{<span class="w">Sequence</span>}{<span class="i">$ID</span>} = <span class="i">RemoveSequenceGaps</span><span class="s">(</span><span class="i">$Sequence</span><span class="s">)</span><span class="sc">;</span>
|
|
131 118 <span class="s">}</span>
|
|
132 119 <span class="s">}</span>
|
|
133 120
|
|
134 121 <span class="c"># Write out the file...</span>
|
|
135 122 <span class="i">WritePearsonFastaSequenceFile</span><span class="s">(</span><span class="i">$OutSequenceFile</span><span class="cm">,</span> <span class="i">$SpecifiedSequenceDataRef</span><span class="cm">,</span> <span class="i">$OptionsInfo</span>{<span class="w">MaxSequenceLength</span>}<span class="s">)</span><span class="sc">;</span>
|
|
136 123 <span class="s">}</span>
|
|
137 124
|
|
138 125 <span class="c"># Get specified sequence data...</span>
|
|
139 <a name="GetSpecifiedSequenceData-"></a> 126 <span class="k">sub </span><span class="m">GetSpecifiedSequenceData</span> <span class="s">{</span>
|
|
140 127 <span class="k">my</span><span class="s">(</span><span class="i">$SequenceDataRef</span><span class="s">)</span> = <span class="i">@_</span><span class="sc">;</span>
|
|
141 128
|
|
142 129 <span class="k">if</span> <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">Mode</span>} =~ <span class="q">/^SequenceID$/i</span><span class="s">)</span> <span class="s">{</span>
|
|
143 130 <span class="k">return</span> <span class="i">GetDataBySequenceIDs</span><span class="s">(</span><span class="i">$SequenceDataRef</span><span class="s">)</span><span class="sc">;</span>
|
|
144 131 <span class="s">}</span>
|
|
145 132 <span class="k">elsif</span> <span class="s">(</span><span class="i">$Options</span>{<span class="w">mode</span>} =~ <span class="q">/^SequenceNum$/i</span><span class="s">)</span> <span class="s">{</span>
|
|
146 133 <span class="k">return</span> <span class="i">GetDataBySequenceNums</span><span class="s">(</span><span class="i">$SequenceDataRef</span><span class="s">)</span><span class="sc">;</span>
|
|
147 134 <span class="s">}</span>
|
|
148 135 <span class="k">elsif</span> <span class="s">(</span><span class="i">$Options</span>{<span class="w">mode</span>} =~ <span class="q">/^SequenceNumRange$/i</span><span class="s">)</span> <span class="s">{</span>
|
|
149 136 <span class="k">return</span> <span class="i">GetDataBySequenceNumRange</span><span class="s">(</span><span class="i">$SequenceDataRef</span><span class="s">)</span><span class="sc">;</span>
|
|
150 137 <span class="s">}</span>
|
|
151 138 <span class="k">else</span> <span class="s">{</span>
|
|
152 139 <span class="k">return</span> <span class="k">undef</span><span class="sc">;</span>
|
|
153 140 <span class="s">}</span>
|
|
154 141 <span class="s">}</span>
|
|
155 142
|
|
156 143 <span class="c"># Get specified sequence data...</span>
|
|
157 <a name="GetDataBySequenceIDs-"></a> 144 <span class="k">sub </span><span class="m">GetDataBySequenceIDs</span> <span class="s">{</span>
|
|
158 145 <span class="k">my</span><span class="s">(</span><span class="i">$SequenceDataRef</span><span class="s">)</span> = <span class="i">@_</span><span class="sc">;</span>
|
|
159 146 <span class="k">my</span><span class="s">(</span><span class="i">$ID</span><span class="cm">,</span> <span class="i">$SequenceCount</span><span class="cm">,</span> <span class="i">$IDMatched</span><span class="cm">,</span> <span class="i">$SpecifiedID</span><span class="cm">,</span> <span class="i">%SpecifiedSequenceDataMap</span><span class="s">)</span><span class="sc">;</span>
|
|
160 147
|
|
161 148 <span class="c"># Go over sequences and collect sequences for writing out a new sequence file...</span>
|
|
162 149 <span class="i">%SpecifiedSequenceDataMap</span> = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
163 150 <span class="i">@</span>{<span class="i">$SpecifiedSequenceDataMap</span>{<span class="w">IDs</span>}} = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
164 151 <span class="i">%</span>{<span class="i">$SpecifiedSequenceDataMap</span>{<span class="w">Description</span>}} = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
165 152 <span class="i">%</span>{<span class="i">$SpecifiedSequenceDataMap</span>{<span class="w">Sequence</span>}} = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
166 153
|
|
167 154 <span class="i">$SequenceCount</span> = <span class="n">0</span><span class="sc">;</span>
|
|
168 155 <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>
|
|
169 156 <span class="k">if</span> <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">MatchExactSequenceIDs</span>}<span class="s">)</span> <span class="s">{</span>
|
|
170 157 <span class="k">if</span> <span class="s">(</span>!<span class="k">exists</span> <span class="i">$OptionsInfo</span>{<span class="w">SpecifiedSequenceIDsMap</span>}{<span class="k">lc</span><span class="s">(</span><span class="i">$ID</span><span class="s">)</span>}<span class="s">)</span> <span class="s">{</span>
|
|
171 158 <span class="k">next</span> <span class="j">ID</span><span class="sc">;</span>
|
|
172 159 <span class="s">}</span>
|
|
173 160 <span class="k">if</span> <span class="s">(</span><span class="i">$SequenceCount</span> >= <span class="k">scalar</span> <span class="i">@</span>{<span class="i">$OptionsInfo</span>{<span class="w">SpecifiedSequenceIDs</span>}}<span class="s">)</span> <span class="s">{</span>
|
|
174 161 <span class="k">last</span> <span class="j">ID</span><span class="sc">;</span>
|
|
175 162 <span class="s">}</span>
|
|
176 163 <span class="i">$SequenceCount</span>++<span class="sc">;</span>
|
|
177 164 <span class="s">}</span>
|
|
178 165 <span class="k">else</span> <span class="s">{</span>
|
|
179 166 <span class="c"># Does this ID contains specified ID as substring...</span>
|
|
180 167 <span class="i">$IDMatched</span> = <span class="n">0</span><span class="sc">;</span>
|
|
181 168 <span class="j">SPECIFIEDID:</span> <span class="k">for</span> <span class="i">$SpecifiedID</span> <span class="s">(</span><span class="i">@</span>{<span class="i">$OptionsInfo</span>{<span class="w">SpecifiedSequenceIDs</span>}}<span class="s">)</span> <span class="s">{</span>
|
|
182 169 <span class="k">if</span> <span class="s">(</span><span class="i">$ID</span> =~ <span class="q">/$SpecifiedID/i</span><span class="s">)</span> <span class="s">{</span>
|
|
183 170 <span class="i">$IDMatched</span> = <span class="n">1</span><span class="sc">;</span>
|
|
184 171 <span class="k">last</span> <span class="j">SPECIFIEDID</span><span class="sc">;</span>
|
|
185 172 <span class="s">}</span>
|
|
186 173 <span class="s">}</span>
|
|
187 174 <span class="k">if</span> <span class="s">(</span>!<span class="i">$IDMatched</span><span class="s">)</span> <span class="s">{</span>
|
|
188 175 <span class="k">next</span> <span class="j">ID</span><span class="sc">;</span>
|
|
189 176 <span class="s">}</span>
|
|
190 177 <span class="i">$SequenceCount</span>++<span class="sc">;</span>
|
|
191 178 <span class="s">}</span>
|
|
192 179 <span class="c"># Collect sequence data...</span>
|
|
193 180 <span class="k">push</span> <span class="i">@</span>{<span class="i">$SpecifiedSequenceDataMap</span>{<span class="w">IDs</span>}}<span class="cm">,</span> <span class="i">$ID</span><span class="sc">;</span>
|
|
194 181 <span class="i">$SpecifiedSequenceDataMap</span>{<span class="w">Description</span>}{<span class="i">$ID</span>} = <span class="i">$SequenceDataRef</span>->{<span class="w">Description</span>}{<span class="i">$ID</span>}<span class="sc">;</span>
|
|
195 182 <span class="i">$SpecifiedSequenceDataMap</span>{<span class="w">Sequence</span>}{<span class="i">$ID</span>} = <span class="i">$SequenceDataRef</span>->{<span class="w">Sequence</span>}{<span class="i">$ID</span>}<span class="sc">;</span>
|
|
196 183 <span class="s">}</span>
|
|
197 184
|
|
198 185 <span class="k">return</span> \<span class="i">%SpecifiedSequenceDataMap</span><span class="sc">;</span>
|
|
199 186 <span class="s">}</span>
|
|
200 187
|
|
201 188 <span class="c"># Get specified sequence data...</span>
|
|
202 <a name="GetDataBySequenceNums-"></a> 189 <span class="k">sub </span><span class="m">GetDataBySequenceNums</span> <span class="s">{</span>
|
|
203 190 <span class="k">my</span><span class="s">(</span><span class="i">$SequenceDataRef</span><span class="s">)</span> = <span class="i">@_</span><span class="sc">;</span>
|
|
204 191 <span class="k">my</span><span class="s">(</span><span class="i">$ID</span><span class="cm">,</span> <span class="i">$SequenceNum</span><span class="cm">,</span> <span class="i">$SequenceCount</span><span class="cm">,</span> <span class="i">%SpecifiedSequenceDataMap</span><span class="s">)</span><span class="sc">;</span>
|
|
205 192
|
|
206 193 <span class="c"># Go over sequences and collect sequences for writing out a new sequence file...</span>
|
|
207 194 <span class="i">%SpecifiedSequenceDataMap</span> = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
208 195 <span class="i">@</span>{<span class="i">$SpecifiedSequenceDataMap</span>{<span class="w">IDs</span>}} = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
209 196 <span class="i">%</span>{<span class="i">$SpecifiedSequenceDataMap</span>{<span class="w">Description</span>}} = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
210 197 <span class="i">%</span>{<span class="i">$SpecifiedSequenceDataMap</span>{<span class="w">Sequence</span>}} = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
211 198
|
|
212 199 <span class="i">$SequenceNum</span> = <span class="n">0</span><span class="sc">;</span>
|
|
213 200 <span class="i">$SequenceCount</span> = <span class="n">0</span><span class="sc">;</span>
|
|
214 201 <span class="j">ID:</span> <span class="k">for</span> <span class="i">$ID</span> <span class="s">(</span><span class="i">@</span>{<span class="i">$SequenceDataRef</span>->{<span class="w">IDs</span>}}<span class="s">)</span> <span class="s">{</span>
|
|
215 202 <span class="i">$SequenceNum</span>++<span class="sc">;</span>
|
|
216 203 <span class="k">if</span> <span class="s">(</span>!<span class="k">exists</span> <span class="i">$OptionsInfo</span>{<span class="w">SpecifiedSequenceIDsMap</span>}{<span class="i">$SequenceNum</span>}<span class="s">)</span> <span class="s">{</span>
|
|
217 204 <span class="k">next</span> <span class="j">ID</span><span class="sc">;</span>
|
|
218 205 <span class="s">}</span>
|
|
219 206 <span class="k">if</span> <span class="s">(</span><span class="i">$SequenceCount</span> >= <span class="k">scalar</span> <span class="i">@</span>{<span class="i">$OptionsInfo</span>{<span class="w">SpecifiedSequenceIDs</span>}}<span class="s">)</span> <span class="s">{</span>
|
|
220 207 <span class="k">last</span> <span class="j">ID</span><span class="sc">;</span>
|
|
221 208 <span class="s">}</span>
|
|
222 209 <span class="i">$SequenceCount</span>++<span class="sc">;</span>
|
|
223 210
|
|
224 211 <span class="c"># Collect sequence data...</span>
|
|
225 212 <span class="k">push</span> <span class="i">@</span>{<span class="i">$SpecifiedSequenceDataMap</span>{<span class="w">IDs</span>}}<span class="cm">,</span> <span class="i">$ID</span><span class="sc">;</span>
|
|
226 213 <span class="i">$SpecifiedSequenceDataMap</span>{<span class="w">Description</span>}{<span class="i">$ID</span>} = <span class="i">$SequenceDataRef</span>->{<span class="w">Description</span>}{<span class="i">$ID</span>}<span class="sc">;</span>
|
|
227 214 <span class="i">$SpecifiedSequenceDataMap</span>{<span class="w">Sequence</span>}{<span class="i">$ID</span>} = <span class="i">$SequenceDataRef</span>->{<span class="w">Sequence</span>}{<span class="i">$ID</span>}<span class="sc">;</span>
|
|
228 215 <span class="s">}</span>
|
|
229 216
|
|
230 217 <span class="k">return</span> \<span class="i">%SpecifiedSequenceDataMap</span><span class="sc">;</span>
|
|
231 218 <span class="s">}</span>
|
|
232 219
|
|
233 220 <span class="c"># Get specified sequence data...</span>
|
|
234 <a name="GetDataBySequenceNumRange-"></a> 221 <span class="k">sub </span><span class="m">GetDataBySequenceNumRange</span> <span class="s">{</span>
|
|
235 222 <span class="k">my</span><span class="s">(</span><span class="i">$SequenceDataRef</span><span class="s">)</span> = <span class="i">@_</span><span class="sc">;</span>
|
|
236 223 <span class="k">my</span><span class="s">(</span><span class="i">$ID</span><span class="cm">,</span> <span class="i">$SequenceNum</span><span class="cm">,</span> <span class="i">$SequenceCount</span><span class="cm">,</span> <span class="i">%SpecifiedSequenceDataMap</span><span class="s">)</span><span class="sc">;</span>
|
|
237 224
|
|
238 225 <span class="c"># Go over sequences and collect sequences for writing out a new sequence file...</span>
|
|
239 226 <span class="i">%SpecifiedSequenceDataMap</span> = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
240 227 <span class="i">@</span>{<span class="i">$SpecifiedSequenceDataMap</span>{<span class="w">IDs</span>}} = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
241 228 <span class="i">%</span>{<span class="i">$SpecifiedSequenceDataMap</span>{<span class="w">Description</span>}} = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
242 229 <span class="i">%</span>{<span class="i">$SpecifiedSequenceDataMap</span>{<span class="w">Sequence</span>}} = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
243 230
|
|
244 231 <span class="i">$SequenceNum</span> = <span class="n">0</span><span class="sc">;</span>
|
|
245 232 <span class="i">$SequenceCount</span> = <span class="n">0</span><span class="sc">;</span>
|
|
246 233 <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>
|
|
247 234 <span class="i">$SequenceNum</span>++<span class="sc">;</span>
|
|
248 235
|
|
249 236 <span class="k">if</span> <span class="s">(</span>!<span class="s">(</span><span class="i">$SequenceNum</span> >= <span class="i">$OptionsInfo</span>{<span class="w">SpecifiedSequenceIDs</span>}[<span class="n">0</span>] && <span class="i">$SequenceNum</span> <= <span class="i">$OptionsInfo</span>{<span class="w">SpecifiedSequenceIDs</span>}[<span class="n">1</span>]<span class="s">)</span><span class="s">)</span> <span class="s">{</span>
|
|
250 237 <span class="k">next</span> <span class="j">ID</span><span class="sc">;</span>
|
|
251 238 <span class="s">}</span>
|
|
252 239 <span class="k">if</span> <span class="s">(</span><span class="i">$SequenceNum</span> > <span class="i">$OptionsInfo</span>{<span class="w">SpecifiedSequenceIDs</span>}[<span class="n">1</span>]<span class="s">)</span> <span class="s">{</span>
|
|
253 240 <span class="k">last</span> <span class="j">ID</span><span class="sc">;</span>
|
|
254 241 <span class="s">}</span>
|
|
255 242 <span class="i">$SequenceCount</span>++<span class="sc">;</span>
|
|
256 243 <span class="c"># Collect sequence data...</span>
|
|
257 244 <span class="k">push</span> <span class="i">@</span>{<span class="i">$SpecifiedSequenceDataMap</span>{<span class="w">IDs</span>}}<span class="cm">,</span> <span class="i">$ID</span><span class="sc">;</span>
|
|
258 245 <span class="i">$SpecifiedSequenceDataMap</span>{<span class="w">Description</span>}{<span class="i">$ID</span>} = <span class="i">$SequenceDataRef</span>->{<span class="w">Description</span>}{<span class="i">$ID</span>}<span class="sc">;</span>
|
|
259 246 <span class="i">$SpecifiedSequenceDataMap</span>{<span class="w">Sequence</span>}{<span class="i">$ID</span>} = <span class="i">$SequenceDataRef</span>->{<span class="w">Sequence</span>}{<span class="i">$ID</span>}<span class="sc">;</span>
|
|
260 247 <span class="s">}</span>
|
|
261 248
|
|
262 249 <span class="k">return</span> \<span class="i">%SpecifiedSequenceDataMap</span><span class="sc">;</span>
|
|
263 250 <span class="s">}</span>
|
|
264 251
|
|
265 252
|
|
266 253 <span class="c"># Process option values...</span>
|
|
267 <a name="ProcessOptions-"></a> 254 <span class="k">sub </span><span class="m">ProcessOptions</span> <span class="s">{</span>
|
|
268 255 <span class="i">%OptionsInfo</span> = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
269 256
|
|
270 257 <span class="c"># Miscellaneous options...</span>
|
|
271 258 <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>
|
|
272 259
|
|
273 260 <span class="i">$OptionsInfo</span>{<span class="w">Mode</span>} = <span class="i">$Options</span>{<span class="w">mode</span>}<span class="sc">;</span>
|
|
274 261 <span class="i">$OptionsInfo</span>{<span class="w">MatchExactSequenceIDs</span>} = <span class="i">$Options</span>{<span class="w">sequenceidmatch</span>} =~ <span class="q">/Exact/i</span> ? <span class="n">1</span> <span class="co">:</span><span class="n">0</span><span class="sc">;</span>
|
|
275 262
|
|
276 263 <span class="c"># Check specified sequences value...</span>
|
|
277 264 <span class="i">$OptionsInfo</span>{<span class="w">SpecifiedSequences</span>} = <span class="i">$Options</span>{<span class="w">sequences</span>}<span class="sc">;</span>
|
|
278 265 <span class="i">@</span>{<span class="i">$OptionsInfo</span>{<span class="w">SpecifiedSequenceIDs</span>}} = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
279 266 <span class="i">%</span>{<span class="i">$OptionsInfo</span>{<span class="w">SpecifiedSequenceIDsMap</span>}} = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
280 267
|
|
281 268 <span class="k">my</span><span class="s">(</span><span class="i">@SpecifiedSequenceIDs</span><span class="s">)</span> = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
282 269 <span class="k">if</span> <span class="s">(</span><span class="i">$Options</span>{<span class="w">mode</span>} =~ <span class="q">/^SequenceID$/i</span><span class="s">)</span> <span class="s">{</span>
|
|
283 270 <span class="k">if</span> <span class="s">(</span>!<span class="i">$Options</span>{<span class="w">sequences</span>}<span class="s">)</span> <span class="s">{</span>
|
|
284 271 <span class="k">die</span> <span class="q">"Error: No value specified for option \"-s, --Sequences\" during \"SequenceID\" of \"-m, --mode\" option\n"</span><span class="sc">;</span>
|
|
285 272 <span class="s">}</span>
|
|
286 273 <span class="i">@SpecifiedSequenceIDs</span> = <span class="k">split</span> <span class="q">/\,/</span><span class="cm">,</span> <span class="i">$Options</span>{<span class="w">sequences</span>}<span class="sc">;</span>
|
|
287 274 <span class="s">}</span>
|
|
288 275 <span class="k">elsif</span> <span class="s">(</span><span class="i">$Options</span>{<span class="w">mode</span>} =~ <span class="q">/^SequenceNum$/i</span><span class="s">)</span> <span class="s">{</span>
|
|
289 276 <span class="k">if</span> <span class="s">(</span><span class="i">$Options</span>{<span class="w">sequences</span>}<span class="s">)</span> <span class="s">{</span>
|
|
290 277 <span class="i">@SpecifiedSequenceIDs</span> = <span class="k">split</span> <span class="q">/\,/</span><span class="cm">,</span> <span class="i">$Options</span>{<span class="w">sequences</span>}<span class="sc">;</span>
|
|
291 278 <span class="k">my</span><span class="s">(</span><span class="i">$SequenceNum</span><span class="s">)</span><span class="sc">;</span>
|
|
292 279 <span class="k">for</span> <span class="i">$SequenceNum</span> <span class="s">(</span><span class="i">@SpecifiedSequenceIDs</span><span class="s">)</span> <span class="s">{</span>
|
|
293 280 <span class="k">if</span> <span class="s">(</span>!<span class="i">IsPositiveInteger</span><span class="s">(</span><span class="i">$SequenceNum</span><span class="s">)</span><span class="s">)</span> <span class="s">{</span>
|
|
294 281 <span class="k">die</span> <span class="q">"Error: The value specified, $SequenceNum, in \"$Options{sequences}\" for option \"-s, --Sequences\" is not valid: Valid values: > 0\n"</span><span class="sc">;</span>
|
|
295 282 <span class="s">}</span>
|
|
296 283 <span class="s">}</span>
|
|
297 284 <span class="s">}</span>
|
|
298 285 <span class="k">else</span> <span class="s">{</span>
|
|
299 286 <span class="k">push</span> <span class="i">@SpecifiedSequenceIDs</span><span class="cm">,</span> <span class="q">"1"</span><span class="sc">;</span>
|
|
300 287 <span class="s">}</span>
|
|
301 288 <span class="s">}</span>
|
|
302 289 <span class="k">elsif</span> <span class="s">(</span><span class="i">$Options</span>{<span class="w">mode</span>} =~ <span class="q">/^SequenceNumRange$/i</span><span class="s">)</span> <span class="s">{</span>
|
|
303 290 <span class="k">if</span> <span class="s">(</span>!<span class="i">$Options</span>{<span class="w">sequences</span>}<span class="s">)</span> <span class="s">{</span>
|
|
304 291 <span class="k">die</span> <span class="q">"Error: No value specified for option \"-s, --Sequences\" during \"SequenceNumRange\" of \"-m, --mode\" option\n"</span><span class="sc">;</span>
|
|
305 292 <span class="s">}</span>
|
|
306 293 <span class="i">@SpecifiedSequenceIDs</span> = <span class="k">split</span> <span class="q">/\,/</span><span class="cm">,</span> <span class="i">$Options</span>{<span class="w">sequences</span>}<span class="sc">;</span>
|
|
307 294 <span class="k">if</span> <span class="s">(</span><span class="i">@SpecifiedSequenceIDs</span> != <span class="n">2</span><span class="s">)</span> <span class="s">{</span>
|
|
308 295 <span class="k">die</span> <span class="q">"Error: The number of values"</span><span class="cm">,</span> <span class="k">scalar</span> <span class="i">@SpecifiedSequenceIDs</span><span class="cm">,</span> <span class="q">" specified, $Options{sequences}, for option \"-s, --Sequences\" are not valid. Number of values must be 2 to indicate starting and ending sequence number.\n"</span><span class="sc">;</span>
|
|
309 296 <span class="s">}</span>
|
|
310 297 <span class="k">my</span><span class="s">(</span><span class="i">$SequenceNum</span><span class="s">)</span><span class="sc">;</span>
|
|
311 298 <span class="k">for</span> <span class="i">$SequenceNum</span> <span class="s">(</span><span class="i">@SpecifiedSequenceIDs</span><span class="s">)</span> <span class="s">{</span>
|
|
312 299 <span class="k">if</span> <span class="s">(</span>!<span class="i">IsPositiveInteger</span><span class="s">(</span><span class="i">$SequenceNum</span><span class="s">)</span><span class="s">)</span> <span class="s">{</span>
|
|
313 300 <span class="k">die</span> <span class="q">"Error: The value specified, $SequenceNum, in \"$Options{sequences}\" for option \"-s, --Sequences\" is not valid: Valid values: > 0\n"</span><span class="sc">;</span>
|
|
314 301 <span class="s">}</span>
|
|
315 302 <span class="s">}</span>
|
|
316 303 <span class="k">if</span> <span class="s">(</span><span class="i">$SpecifiedSequenceIDs</span>[<span class="n">0</span>] > <span class="i">$SpecifiedSequenceIDs</span>[<span class="n">1</span>]<span class="s">)</span> <span class="s">{</span>
|
|
317 304 <span class="k">die</span> <span class="q">"Error: The value specified \"$Options{sequences}\" for option \"-s, --Sequences\" are not valid: Starting sequence number $SpecifiedSequenceIDs[0] must be smaller than ending sequence number $SpecifiedSequenceIDs[1]\n"</span><span class="sc">;</span>
|
|
318 305 <span class="s">}</span>
|
|
319 306 <span class="s">}</span>
|
|
320 307 <span class="k">push</span> <span class="i">@</span>{<span class="i">$OptionsInfo</span>{<span class="w">SpecifiedSequenceIDs</span>}}<span class="cm">,</span> <span class="i">@SpecifiedSequenceIDs</span><span class="sc">;</span>
|
|
321 308 <span class="k">my</span><span class="s">(</span><span class="i">$SequenceID</span><span class="s">)</span><span class="sc">;</span>
|
|
322 309 <span class="k">for</span> <span class="i">$SequenceID</span> <span class="s">(</span><span class="i">@SpecifiedSequenceIDs</span><span class="s">)</span> <span class="s">{</span>
|
|
323 310 <span class="k">if</span> <span class="s">(</span><span class="i">$Options</span>{<span class="w">mode</span>} =~ <span class="q">/^SequenceID$/i</span><span class="s">)</span> <span class="s">{</span>
|
|
324 311 <span class="i">$OptionsInfo</span>{<span class="w">SpecifiedSequenceIDsMap</span>}{<span class="k">lc</span><span class="s">(</span><span class="i">$SequenceID</span><span class="s">)</span>} = <span class="i">$SequenceID</span><span class="sc">;</span>
|
|
325 312 <span class="s">}</span>
|
|
326 313 <span class="k">else</span> <span class="s">{</span>
|
|
327 314 <span class="i">$OptionsInfo</span>{<span class="w">SpecifiedSequenceIDsMap</span>}{<span class="i">$SequenceID</span>} = <span class="i">$SequenceID</span><span class="sc">;</span>
|
|
328 315 <span class="s">}</span>
|
|
329 316 <span class="s">}</span>
|
|
330 317
|
|
331 318 <span class="i">$OptionsInfo</span>{<span class="w">MaxSequenceLength</span>} = <span class="i">$Options</span>{<span class="w">sequencelength</span>}<span class="sc">;</span>
|
|
332 319 <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>
|
|
333 320 <span class="i">$OptionsInfo</span>{<span class="w">OutFileRoot</span>} = <span class="i">$Options</span>{<span class="w">root</span>} ? <span class="i">$Options</span>{<span class="w">root</span>} <span class="co">:</span> <span class="n">0</span><span class="sc">;</span>
|
|
334 321 <span class="s">}</span>
|
|
335 322
|
|
336 323 <span class="c"># Retrieve information about sequence files...</span>
|
|
337 <a name="RetrieveSequenceFilesInfo-"></a> 324 <span class="k">sub </span><span class="m">RetrieveSequenceFilesInfo</span> <span class="s">{</span>
|
|
338 325 <span class="k">my</span><span class="s">(</span><span class="i">$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">$FileDir</span><span class="cm">,</span> <span class="i">$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">$OutFileExt</span><span class="cm">,</span> <span class="i">$OutFileMode</span><span class="cm">,</span> <span class="i">$SequenceDataRef</span><span class="s">)</span><span class="sc">;</span>
|
|
339 326
|
|
340 327 <span class="i">%SequenceFilesInfo</span> = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
341 328 <span class="i">@</span>{<span class="i">$SequenceFilesInfo</span>{<span class="w">FilesOkay</span>}} = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
342 329 <span class="i">@</span>{<span class="i">$SequenceFilesInfo</span>{<span class="w">OutFileRoot</span>}} = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
343 330 <span class="i">@</span>{<span class="i">$SequenceFilesInfo</span>{<span class="w">OutFileExt</span>}} = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
344 331 <span class="i">@</span>{<span class="i">$SequenceFilesInfo</span>{<span class="w">OutFile</span>}} = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
345 332 <span class="i">@</span>{<span class="i">$SequenceFilesInfo</span>{<span class="w">Format</span>}} = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
346 333 <span class="i">@</span>{<span class="i">$SequenceFilesInfo</span>{<span class="w">SequenceCount</span>}} = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
347 334
|
|
348 335 <span class="j">FILELIST:</span> <span class="k">for</span> <span class="i">$Index</span> <span class="s">(</span><span class="n">0</span> .. <span class="i">$#SequenceFilesList</span><span class="s">)</span> <span class="s">{</span>
|
|
349 336 <span class="i">$SequenceFile</span> = <span class="i">$SequenceFilesList</span>[<span class="i">$Index</span>]<span class="sc">;</span>
|
|
350 337 <span class="i">$SequenceFilesInfo</span>{<span class="w">FilesOkay</span>}[<span class="i">$Index</span>] = <span class="n">0</span><span class="sc">;</span>
|
|
351 338 <span class="i">$SequenceFilesInfo</span>{<span class="w">OutFileRoot</span>}[<span class="i">$Index</span>] = <span class="q">''</span><span class="sc">;</span>
|
|
352 339 <span class="i">$SequenceFilesInfo</span>{<span class="w">OutFileExt</span>}[<span class="i">$Index</span>] = <span class="q">''</span><span class="sc">;</span>
|
|
353 340 <span class="i">$SequenceFilesInfo</span>{<span class="w">OutFile</span>}[<span class="i">$Index</span>] = <span class="q">''</span><span class="sc">;</span>
|
|
354 341 <span class="i">$SequenceFilesInfo</span>{<span class="w">Format</span>}[<span class="i">$Index</span>] = <span class="q">'NotSupported'</span><span class="sc">;</span>
|
|
355 342 <span class="i">$SequenceFilesInfo</span>{<span class="w">SequenceCount</span>}[<span class="i">$Index</span>] = <span class="n">0</span><span class="sc">;</span>
|
|
356 343
|
|
357 344 <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>
|
|
358 345 <span class="k">warn</span> <span class="q">"Warning: Ignoring file $SequenceFile: Couldn't open it: $! \n"</span><span class="sc">;</span>
|
|
359 346 <span class="k">next</span> <span class="j">FILELIST</span><span class="sc">;</span>
|
|
360 347 <span class="s">}</span>
|
|
361 348 <span class="k">close</span> <span class="w">SEQUENCEFILE</span><span class="sc">;</span>
|
|
362 349
|
|
363 350 <span class="s">(</span><span class="i">$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>
|
|
364 351 <span class="k">if</span> <span class="s">(</span>!<span class="i">$FileSupported</span><span class="s">)</span> <span class="s">{</span>
|
|
365 352 <span class="k">warn</span> <span class="q">"Warning: Ignoring file $SequenceFile: Sequence file format is not supported.\n"</span><span class="sc">;</span>
|
|
366 353 <span class="k">next</span> <span class="j">FILELIST</span><span class="sc">;</span>
|
|
367 354 <span class="s">}</span>
|
|
368 355 <span class="i">$SequenceDataRef</span> = <span class="i">ReadSequenceFile</span><span class="s">(</span><span class="i">$SequenceFile</span><span class="s">)</span><span class="sc">;</span>
|
|
369 356
|
|
370 357 <span class="i">$SequenceCount</span> = <span class="i">$SequenceDataRef</span>->{<span class="w">Count</span>}<span class="sc">;</span>
|
|
371 358 <span class="k">if</span> <span class="s">(</span>!<span class="i">$SequenceCount</span><span class="s">)</span> <span class="s">{</span>
|
|
372 359 <span class="k">warn</span> <span class="q">"Warning: Ignoring file $SequenceFile: Sequence data is missing.\n"</span><span class="sc">;</span>
|
|
373 360 <span class="k">next</span> <span class="j">FILELIST</span><span class="sc">;</span>
|
|
374 361 <span class="s">}</span>
|
|
375 362
|
|
376 363 <span class="c"># Setup output file names...</span>
|
|
377 364 <span class="i">$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>
|
|
378 365 <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>
|
|
379 366 <span class="i">$OutFileExt</span> = <span class="q">'fasta'</span><span class="sc">;</span>
|
|
380 367 <span class="k">if</span> <span class="s">(</span><span class="i">$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>
|
|
381 368 <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>
|
|
382 369 <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>
|
|
383 370 <span class="i">$FileName</span> = <span class="i">$RootFileName</span><span class="sc">;</span>
|
|
384 371 <span class="s">}</span>
|
|
385 372 <span class="k">else</span> <span class="s">{</span>
|
|
386 373 <span class="i">$FileName</span> = <span class="i">$OptionsInfo</span>{<span class="w">OutFileRoot</span>}<span class="sc">;</span>
|
|
387 374 <span class="s">}</span>
|
|
388 375 <span class="i">$OutFileRoot</span> = <span class="i">$FileName</span><span class="sc">;</span>
|
|
389 376 <span class="s">}</span>
|
|
390 377 <span class="k">else</span> <span class="s">{</span>
|
|
391 378 <span class="i">$OutFileRoot</span> = <span class="i">$FileName</span><span class="sc">;</span>
|
|
392 379 <span class="s">}</span>
|
|
393 380 <span class="j">MODE:</span> <span class="s">{</span>
|
|
394 381 <span class="k">if</span> <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">Mode</span>} =~ <span class="q">/^SequenceID$/i</span><span class="s">)</span> <span class="s">{</span> <span class="i">$OutFileMode</span> = <span class="q">'SequenceID'</span><span class="sc">;</span> <span class="k">last</span> <span class="j">MODE</span><span class="sc">;</span><span class="s">}</span>
|
|
395 382 <span class="k">if</span> <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">Mode</span>} =~ <span class="q">/^SequenceNum$/i</span><span class="s">)</span> <span class="s">{</span> <span class="i">$OutFileMode</span> = <span class="q">'SequenceNum'</span><span class="sc">;</span> <span class="k">last</span> <span class="j">MODE</span><span class="sc">;</span><span class="s">}</span>
|
|
396 383 <span class="k">if</span> <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">Mode</span>} =~ <span class="q">/^SequenceNumRange$/i</span><span class="s">)</span> <span class="s">{</span> <span class="i">$OutFileMode</span> = <span class="q">'SequenceNumRange'</span><span class="sc">;</span> <span class="k">last</span> <span class="j">MODE</span><span class="sc">;</span><span class="s">}</span>
|
|
397 384 <span class="i">$OutFileMode</span> = <span class="q">''</span><span class="sc">;</span>
|
|
398 385 <span class="s">}</span>
|
|
399 386 <span class="k">if</span> <span class="s">(</span>!<span class="i">$OptionsInfo</span>{<span class="w">OverwriteFiles</span>}<span class="s">)</span> <span class="s">{</span>
|
|
400 387 <span class="k">if</span> <span class="s">(</span><span class="k">-e</span> <span class="q">"${OutFileRoot}${OutFileMode}.${OutFileExt}"</span><span class="s">)</span> <span class="s">{</span>
|
|
401 388 <span class="k">warn</span> <span class="q">"Warning: Ignoring file $SequenceFile: The file ${OutFileRoot}${OutFileMode}.${OutFileExt} already exists\n"</span><span class="sc">;</span>
|
|
402 389 <span class="k">next</span> <span class="j">FILELIST</span><span class="sc">;</span>
|
|
403 390 <span class="s">}</span>
|
|
404 391 <span class="s">}</span>
|
|
405 392
|
|
406 393 <span class="i">$SequenceFilesInfo</span>{<span class="w">FilesOkay</span>}[<span class="i">$Index</span>] = <span class="n">1</span><span class="sc">;</span>
|
|
407 394 <span class="i">$SequenceFilesInfo</span>{<span class="w">OutFileRoot</span>}[<span class="i">$Index</span>] = <span class="i">$OutFileRoot</span><span class="sc">;</span>
|
|
408 395 <span class="i">$SequenceFilesInfo</span>{<span class="w">OutFileExt</span>}[<span class="i">$Index</span>] = <span class="i">$OutFileExt</span><span class="sc">;</span>
|
|
409 396 <span class="i">$SequenceFilesInfo</span>{<span class="w">OutFile</span>}[<span class="i">$Index</span>] = <span class="q">"${OutFileRoot}${OutFileMode}.${OutFileExt}"</span><span class="sc">;</span>
|
|
410 397
|
|
411 398 <span class="i">$SequenceFilesInfo</span>{<span class="w">Format</span>}[<span class="i">$Index</span>] = <span class="i">$FileFormat</span><span class="sc">;</span>
|
|
412 399 <span class="i">$SequenceFilesInfo</span>{<span class="w">SequenceCount</span>}[<span class="i">$Index</span>] = <span class="i">$SequenceCount</span><span class="sc">;</span>
|
|
413 400 <span class="s">}</span>
|
|
414 401 <span class="s">}</span>
|
|
415 402
|
|
416 403 <span class="c"># Setup script usage and retrieve command line arguments specified using various options...</span>
|
|
417 <a name="SetupScriptUsage-"></a> 404 <span class="k">sub </span><span class="m">SetupScriptUsage</span> <span class="s">{</span>
|
|
418 405
|
|
419 406 <span class="c"># Retrieve all the options...</span>
|
|
420 407 <span class="i">%Options</span> = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
421 408 <span class="i">$Options</span>{<span class="w">ignoregaps</span>} = <span class="q">'Yes'</span><span class="sc">;</span>
|
|
422 409 <span class="i">$Options</span>{<span class="w">mode</span>} = <span class="q">'SequenceNum'</span><span class="sc">;</span>
|
|
423 410 <span class="i">$Options</span>{<span class="w">sequenceidmatch</span>} = <span class="q">'Relaxed'</span><span class="sc">;</span>
|
|
424 411 <span class="i">$Options</span>{<span class="w">sequencelength</span>} = <span class="n">80</span><span class="sc">;</span>
|
|
425 412
|
|
426 413 <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">"overwrite|o"</span><span class="cm">,</span> <span class="q">"root|r=s"</span><span class="cm">,</span> <span class="q">"sequences|s=s"</span><span class="cm">,</span> <span class="q">"sequenceidmatch=s"</span><span class="cm">,</span> <span class="q">"sequencelength=i"</span><span class="cm">,</span> <span class="q">"workingdir|w=s"</span><span class="s">)</span><span class="s">)</span> <span class="s">{</span>
|
|
427 414 <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 $ScriptName -h\" command and try again...\n"</span><span class="sc">;</span>
|
|
428 415 <span class="s">}</span>
|
|
429 416 <span class="k">if</span> <span class="s">(</span><span class="i">$Options</span>{<span class="w">workingdir</span>}<span class="s">)</span> <span class="s">{</span>
|
|
430 417 <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>
|
|
431 418 <span class="k">die</span> <span class="q">"Error: The value specified, $Options{workingdir}, for option \"-w --workingdir\" is not a directory name.\n"</span><span class="sc">;</span>
|
|
432 419 <span class="s">}</span>
|
|
433 420 <span class="k">chdir</span> <span class="i">$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}: $! \n"</span><span class="sc">;</span>
|
|
434 421 <span class="s">}</span>
|
|
435 422 <span class="k">if</span> <span class="s">(</span><span class="i">$Options</span>{<span class="w">ignoregaps</span>} !~ <span class="q">/^(yes|no)$/i</span><span class="s">)</span> <span class="s">{</span>
|
|
436 423 <span class="k">die</span> <span class="q">"Error: The value specified, $Options{ignoregaps}, for option \"-q --quote\" is not valid. Allowed values: yes or no\n"</span><span class="sc">;</span>
|
|
437 424 <span class="s">}</span>
|
|
438 425 <span class="k">if</span> <span class="s">(</span><span class="i">$Options</span>{<span class="w">mode</span>} !~ <span class="q">/^(SequenceID|SequenceNum|SequenceNumRange)$/i</span><span class="s">)</span> <span class="s">{</span>
|
|
439 426 <span class="k">die</span> <span class="q">"Error: The value specified, $Options{mode}, for option \"-m --mode\" is not valid. Allowed values: SequenceID, SequenceNum, or SequenceNumRange\n"</span><span class="sc">;</span>
|
|
440 427 <span class="s">}</span>
|
|
441 428 <span class="k">if</span> <span class="s">(</span><span class="i">$Options</span>{<span class="w">sequenceidmatch</span>} !~ <span class="q">/^(Exact|Relaxed)$/i</span><span class="s">)</span> <span class="s">{</span>
|
|
442 429 <span class="k">die</span> <span class="q">"Error: The value specified, $Options{sequenceidmatch}, for option \"--SequenceIDMatch\" is not valid. Allowed values: Exact or Relaxed\n"</span><span class="sc">;</span>
|
|
443 430 <span class="s">}</span>
|
|
444 431 <span class="k">if</span> <span class="s">(</span>!<span class="i">IsPositiveInteger</span><span class="s">(</span><span class="i">$Options</span>{<span class="w">sequencelength</span>}<span class="s">)</span><span class="s">)</span> <span class="s">{</span>
|
|
445 432 <span class="k">die</span> <span class="q">"Error: The value specified, $Options{sequencelength}, for option \"--SequenceLength\" is not valid. Allowed values: >0\n"</span><span class="sc">;</span>
|
|
446 433 <span class="s">}</span>
|
|
447 434 <span class="s">}</span>
|
|
448 435
|
|
449 <a name="EOF-"></a></pre>
|
|
450 <p> </p>
|
|
451 <br />
|
|
452 <center>
|
|
453 <img src="../../../images/h2o2.png">
|
|
454 </center>
|
|
455 </body>
|
|
456 </html>
|