comparison docs/scripts/html/code/InfoSequenceFiles.html @ 0:4816e4a8ae95 draft default tip

Uploaded
author deepakjadmin
date Wed, 20 Jan 2016 09:23:18 -0500
parents
children
comparison
equal deleted inserted replaced
-1:000000000000 0:4816e4a8ae95
1 <html>
2 <head>
3 <title>MayaChemTools:Code:InfoSequenceFiles.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: InfoSequenceFiles.pl,v $</span>
17 4 <span class="c"># $Date: 2015/02/28 20:46:20 $</span>
18 5 <span class="c"># $Revision: 1.29 $</span>
19 6 <span class="c">#</span>
20 7 <span class="c"># Author: Manish Sud &lt;msud@san.rr.com&gt;</span>
21 8 <span class="c">#</span>
22 9 <span class="c"># Copyright (C) 2015 Manish Sud. All rights reserved.</span>
23 10 <span class="c">#</span>
24 11 <span class="c"># This file is part of MayaChemTools.</span>
25 12 <span class="c">#</span>
26 13 <span class="c"># MayaChemTools is free software; you can redistribute it and/or modify it under</span>
27 14 <span class="c"># the terms of the GNU Lesser General Public License as published by the Free</span>
28 15 <span class="c"># Software Foundation; either version 3 of the License, or (at your option) any</span>
29 16 <span class="c"># later version.</span>
30 17 <span class="c">#</span>
31 18 <span class="c"># MayaChemTools is distributed in the hope that it will be useful, but without</span>
32 19 <span class="c"># any warranty; without even the implied warranty of merchantability of fitness</span>
33 20 <span class="c"># for a particular purpose. See the GNU Lesser General Public License for more</span>
34 21 <span class="c"># details.</span>
35 22 <span class="c">#</span>
36 23 <span class="c"># You should have received a copy of the GNU Lesser General Public License</span>
37 24 <span class="c"># along with MayaChemTools; if not, see &lt;http://www.gnu.org/licenses/&gt; or</span>
38 25 <span class="c"># write to the Free Software Foundation Inc., 59 Temple Place, Suite 330,</span>
39 26 <span class="c"># Boston, MA, 02111-1307, USA.</span>
40 27 <span class="c">#</span>
41 28
42 29 <span class="k">use</span> <span class="w">strict</span><span class="sc">;</span>
43 30 <span class="k">use</span> <span class="w">FindBin</span><span class="sc">;</span> <span class="k">use</span> <span class="w">lib</span> <span class="q">&quot;$FindBin::Bin/../lib&quot;</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">StatisticsUtil</span><span class="sc">;</span>
52 39
53 40 <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>
54 41
55 42 <span class="c"># Autoflush STDOUT</span>
56 43 <span class="i">$|</span> = <span class="n">1</span><span class="sc">;</span>
57 44
58 45 <span class="c"># Starting message...</span>
59 46 <span class="i">$ScriptName</span> = <span class="i">basename</span><span class="s">(</span><span class="i">$0</span><span class="s">)</span><span class="sc">;</span>
60 47 <span class="k">print</span> <span class="q">&quot;\n$ScriptName: Starting...\n\n&quot;</span><span class="sc">;</span>
61 48 <span class="i">$StartTime</span> = <span class="w">new</span> <span class="w">Benchmark</span><span class="sc">;</span>
62 49
63 50 <span class="c"># Get the options and setup script...</span>
64 51 <span class="i">SetupScriptUsage</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span>
65 52 <span class="k">if</span> <span class="s">(</span><span class="i">$Options</span>{<span class="w">help</span>} || <span class="i">@ARGV</span> &lt; <span class="n">1</span><span class="s">)</span> <span class="s">{</span>
66 53 <span class="k">die</span> <span class="i">GetUsageFromPod</span><span class="s">(</span><span class="q">&quot;$FindBin::Bin/$ScriptName&quot;</span><span class="s">)</span><span class="sc">;</span>
67 54 <span class="s">}</span>
68 55
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">&quot;aln msf fasta fta pir&quot;</span><span class="s">)</span><span class="sc">;</span>
71 58
72 59 <span class="k">print</span> <span class="q">&quot;Processing options...\n&quot;</span><span class="sc">;</span>
73 60 <span class="k">my</span><span class="s">(</span><span class="i">%OptionsInfo</span><span class="s">)</span><span class="sc">;</span>
74 61 <span class="i">ProcessOptions</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span>
75 62
76 63 <span class="k">print</span> <span class="q">&quot;Checking input sequence file(s)...\n&quot;</span><span class="sc">;</span>
77 64 <span class="k">my</span><span class="s">(</span><span class="i">%SequenceFilesInfo</span><span class="s">)</span><span class="sc">;</span>
78 65 <span class="i">RetrieveSequenceFilesInfo</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span>
79 66
80 67 <span class="k">my</span><span class="s">(</span><span class="i">$FileIndex</span><span class="s">)</span><span class="sc">;</span>
81 68 <span class="k">if</span> <span class="s">(</span><span class="i">@SequenceFilesList</span> &gt; <span class="n">1</span><span class="s">)</span> <span class="s">{</span>
82 69 <span class="k">print</span> <span class="q">&quot;\nProcessing sequence files...\n&quot;</span><span class="sc">;</span>
83 70 <span class="s">}</span>
84 71 <span class="k">for</span> <span class="i">$FileIndex</span> <span class="s">(</span><span class="n">0</span> .. <span class="i">$#SequenceFilesList</span><span class="s">)</span> <span class="s">{</span>
85 72 <span class="k">if</span> <span class="s">(</span><span class="i">$SequenceFilesInfo</span>{<span class="w">FileOkay</span>}[<span class="i">$FileIndex</span>]<span class="s">)</span> <span class="s">{</span>
86 73 <span class="k">print</span> <span class="q">&quot;\nProcessing file $SequenceFilesList[$FileIndex]...\n&quot;</span><span class="sc">;</span>
87 74 <span class="i">ListSequenceFileInfo</span><span class="s">(</span><span class="i">$FileIndex</span><span class="s">)</span><span class="sc">;</span>
88 75 <span class="s">}</span>
89 76 <span class="s">}</span>
90 77 <span class="i">ListTotalSizeOfFiles</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span>
91 78
92 79 <span class="k">print</span> <span class="q">&quot;\n$ScriptName:Done...\n\n&quot;</span><span class="sc">;</span>
93 80
94 81 <span class="i">$EndTime</span> = <span class="w">new</span> <span class="w">Benchmark</span><span class="sc">;</span>
95 82 <span class="i">$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>
96 83 <span class="k">print</span> <span class="q">&quot;Total time: &quot;</span><span class="cm">,</span> <span class="i">timestr</span><span class="s">(</span><span class="i">$TotalTime</span><span class="s">)</span><span class="cm">,</span> <span class="q">&quot;\n&quot;</span><span class="sc">;</span>
97 84
98 85 <span class="c">###############################################################################</span>
99 86
100 87 <span class="c"># List appropriate information...</span>
101 <a name="ListSequenceFileInfo-"></a> 88 <span class="k">sub </span><span class="m">ListSequenceFileInfo</span> <span class="s">{</span>
102 89 <span class="k">my</span><span class="s">(</span><span class="i">$Index</span><span class="s">)</span> = <span class="i">@_</span><span class="sc">;</span>
103 90 <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>
104 91
105 92 <span class="i">$SequenceFile</span> = <span class="i">$SequenceFilesList</span>[<span class="i">$Index</span>]<span class="sc">;</span>
106 93
107 94 <span class="i">$SequenceDataRef</span> = <span class="i">ReadSequenceFile</span><span class="s">(</span><span class="i">$SequenceFile</span><span class="s">)</span><span class="sc">;</span>
108 95
109 96 <span class="k">my</span><span class="s">(</span><span class="i">$SequencesCount</span><span class="s">)</span> = <span class="i">$SequenceDataRef</span>-&gt;{<span class="w">Count</span>}<span class="sc">;</span>
110 97 <span class="k">print</span> <span class="q">&quot;\nNumber of sequences: $SequencesCount\n&quot;</span><span class="sc">;</span>
111 98
112 99 <span class="k">if</span> <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">ListShortestSequence</span>} &amp;&amp; <span class="s">(</span><span class="i">$SequencesCount</span> &gt; <span class="n">1</span><span class="s">)</span><span class="s">)</span> <span class="s">{</span>
113 100 <span class="k">my</span><span class="s">(</span><span class="i">$ShortestSeqID</span><span class="cm">,</span> <span class="i">$ShortestSeq</span><span class="cm">,</span> <span class="i">$ShortestSeqLen</span><span class="cm">,</span> <span class="i">$Description</span><span class="s">)</span> = <span class="i">GetShortestSequence</span><span class="s">(</span><span class="i">$SequenceDataRef</span><span class="cm">,</span> <span class="i">$OptionsInfo</span>{<span class="w">IgnoreGaps</span>}<span class="s">)</span><span class="sc">;</span>
114 101 <span class="k">print</span> <span class="q">&quot;\nShortest sequence information:\nID: $ShortestSeqID; Length:$ShortestSeqLen\n&quot;</span><span class="sc">;</span>
115 102 <span class="k">if</span> <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">DetailLevel</span>} &gt;= <span class="n">2</span><span class="s">)</span> <span class="s">{</span>
116 103 <span class="k">print</span> <span class="q">&quot;Description: $Description\n&quot;</span><span class="sc">;</span>
117 104 <span class="s">}</span>
118 105 <span class="k">if</span> <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">DetailLevel</span>} &gt;= <span class="n">3</span><span class="s">)</span> <span class="s">{</span>
119 106 <span class="k">print</span> <span class="q">&quot;Sequence: $ShortestSeq\n&quot;</span><span class="sc">;</span>
120 107 <span class="s">}</span>
121 108 <span class="s">}</span>
122 109 <span class="k">if</span> <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">ListLongestSequence</span>} &amp;&amp; <span class="s">(</span><span class="i">$SequencesCount</span> &gt; <span class="n">1</span><span class="s">)</span><span class="s">)</span> <span class="s">{</span>
123 110 <span class="k">my</span><span class="s">(</span><span class="i">$LongestSeqID</span><span class="cm">,</span> <span class="i">$LongestSeq</span><span class="cm">,</span> <span class="i">$LongestSeqLen</span><span class="cm">,</span> <span class="i">$Description</span><span class="s">)</span> = <span class="i">GetLongestSequence</span><span class="s">(</span><span class="i">$SequenceDataRef</span><span class="cm">,</span> <span class="i">$OptionsInfo</span>{<span class="w">IgnoreGaps</span>}<span class="s">)</span><span class="sc">;</span>
124 111 <span class="k">print</span> <span class="q">&quot;\nLongest sequence information:\nID: $LongestSeqID; Length: $LongestSeqLen\n&quot;</span><span class="sc">;</span>
125 112 <span class="k">if</span> <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">DetailLevel</span>} &gt;= <span class="n">2</span><span class="s">)</span> <span class="s">{</span>
126 113 <span class="k">print</span> <span class="q">&quot;Description: $Description\n&quot;</span><span class="sc">;</span>
127 114 <span class="s">}</span>
128 115 <span class="k">if</span> <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">DetailLevel</span>} &gt;= <span class="n">3</span><span class="s">)</span> <span class="s">{</span>
129 116 <span class="k">print</span> <span class="q">&quot;Sequence: $LongestSeq\n&quot;</span><span class="sc">;</span>
130 117 <span class="s">}</span>
131 118 <span class="s">}</span>
132 119 <span class="k">if</span> <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">FrequencyAnalysis</span>} &amp;&amp; <span class="s">(</span><span class="i">$SequencesCount</span> &gt; <span class="n">1</span><span class="s">)</span><span class="s">)</span> <span class="s">{</span>
133 120 <span class="i">PerformLengthFrequencyAnalysis</span><span class="s">(</span><span class="i">$SequenceDataRef</span><span class="s">)</span><span class="sc">;</span>
134 121 <span class="s">}</span>
135 122 <span class="k">if</span> <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">ListSequenceLengths</span>}<span class="s">)</span> <span class="s">{</span>
136 123 <span class="i">ListSequenceLengths</span><span class="s">(</span><span class="i">$SequenceDataRef</span><span class="s">)</span><span class="sc">;</span>
137 124 <span class="s">}</span>
138 125
139 126 <span class="c"># File size and modification information...</span>
140 127 <span class="k">print</span> <span class="q">&quot;\nFile size: &quot;</span><span class="cm">,</span> <span class="i">FormatFileSize</span><span class="s">(</span><span class="i">$SequenceFilesInfo</span>{<span class="w">FileSize</span>}[<span class="i">$Index</span>]<span class="s">)</span><span class="cm">,</span> <span class="q">&quot; \n&quot;</span><span class="sc">;</span>
141 128 <span class="k">print</span> <span class="q">&quot;Last modified: &quot;</span><span class="cm">,</span> <span class="i">$SequenceFilesInfo</span>{<span class="w">FileLastModified</span>}[<span class="i">$Index</span>]<span class="cm">,</span> <span class="q">&quot; \n&quot;</span><span class="sc">;</span>
142 129 <span class="s">}</span>
143 130
144 131 <span class="c"># List information about sequence lengths...</span>
145 <a name="ListSequenceLengths-"></a> 132 <span class="k">sub </span><span class="m">ListSequenceLengths</span> <span class="s">{</span>
146 133 <span class="k">my</span><span class="s">(</span><span class="i">$SequenceDataRef</span><span class="s">)</span> = <span class="i">@_</span><span class="sc">;</span>
147 134 <span class="k">my</span><span class="s">(</span><span class="i">$ID</span><span class="cm">,</span> <span class="i">$SeqLen</span><span class="cm">,</span> <span class="i">$Sequence</span><span class="cm">,</span> <span class="i">$Description</span><span class="s">)</span><span class="sc">;</span>
148 135
149 136 <span class="k">print</span> <span class="q">&quot;\nSequence lengths information:\n&quot;</span><span class="sc">;</span>
150 137 <span class="k">for</span> <span class="i">$ID</span> <span class="s">(</span><span class="i">@</span>{<span class="i">$SequenceDataRef</span>-&gt;{<span class="w">IDs</span>}}<span class="s">)</span> <span class="s">{</span>
151 138 <span class="i">$Sequence</span> = <span class="i">$SequenceDataRef</span>-&gt;{<span class="w">Sequence</span>}{<span class="i">$ID</span>}<span class="sc">;</span>
152 139 <span class="i">$Description</span> = <span class="i">$SequenceDataRef</span>-&gt;{<span class="w">Description</span>}{<span class="i">$ID</span>}<span class="sc">;</span>
153 140 <span class="i">$SeqLen</span> = <span class="i">GetSequenceLength</span><span class="s">(</span><span class="i">$Sequence</span><span class="cm">,</span> <span class="i">$OptionsInfo</span>{<span class="w">IgnoreGaps</span>}<span class="s">)</span><span class="sc">;</span>
154 141 <span class="k">if</span> <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">IgnoreGaps</span>}<span class="s">)</span> <span class="s">{</span>
155 142 <span class="i">$Sequence</span> = <span class="i">RemoveSequenceGaps</span><span class="s">(</span><span class="i">$Sequence</span><span class="s">)</span><span class="sc">;</span>
156 143 <span class="s">}</span>
157 144 <span class="k">print</span> <span class="q">&quot;ID: $ID; Length:$SeqLen\n&quot;</span><span class="sc">;</span>
158 145 <span class="k">if</span> <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">DetailLevel</span>} &gt;= <span class="n">2</span><span class="s">)</span> <span class="s">{</span>
159 146 <span class="k">print</span> <span class="q">&quot;Description: $Description\n&quot;</span><span class="sc">;</span>
160 147 <span class="s">}</span>
161 148 <span class="k">if</span> <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">DetailLevel</span>} &gt;= <span class="n">3</span><span class="s">)</span> <span class="s">{</span>
162 149 <span class="k">print</span> <span class="q">&quot;Sequence: $Sequence\n&quot;</span><span class="sc">;</span>
163 150 <span class="s">}</span>
164 151 <span class="k">if</span> <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">DetailLevel</span>} &gt;= <span class="n">2</span><span class="s">)</span> <span class="s">{</span>
165 152 <span class="k">print</span> <span class="q">&quot;\n&quot;</span><span class="sc">;</span>
166 153 <span class="s">}</span>
167 154 <span class="s">}</span>
168 155 <span class="s">}</span>
169 156
170 157 <span class="c"># Total size of all the fiels...</span>
171 <a name="ListTotalSizeOfFiles-"></a> 158 <span class="k">sub </span><span class="m">ListTotalSizeOfFiles</span> <span class="s">{</span>
172 159 <span class="k">my</span><span class="s">(</span><span class="i">$FileOkayCount</span><span class="cm">,</span> <span class="i">$TotalSize</span><span class="cm">,</span> <span class="i">$Index</span><span class="s">)</span><span class="sc">;</span>
173 160
174 161 <span class="i">$FileOkayCount</span> = <span class="n">0</span><span class="sc">;</span>
175 162 <span class="i">$TotalSize</span> = <span class="n">0</span><span class="sc">;</span>
176 163
177 164 <span class="k">for</span> <span class="i">$Index</span> <span class="s">(</span><span class="n">0</span> .. <span class="i">$#SequenceFilesList</span><span class="s">)</span> <span class="s">{</span>
178 165 <span class="k">if</span> <span class="s">(</span><span class="i">$SequenceFilesInfo</span>{<span class="w">FileOkay</span>}[<span class="i">$Index</span>]<span class="s">)</span> <span class="s">{</span>
179 166 <span class="i">$FileOkayCount</span>++<span class="sc">;</span>
180 167 <span class="i">$TotalSize</span> += <span class="i">$SequenceFilesInfo</span>{<span class="w">FileSize</span>}[<span class="i">$Index</span>]<span class="sc">;</span>
181 168 <span class="s">}</span>
182 169 <span class="s">}</span>
183 170 <span class="k">if</span> <span class="s">(</span><span class="i">$FileOkayCount</span> &gt; <span class="n">1</span><span class="s">)</span> <span class="s">{</span>
184 171 <span class="k">print</span> <span class="q">&quot;\nTotal size of $FileOkayCount files: &quot;</span><span class="cm">,</span> <span class="i">FormatFileSize</span><span class="s">(</span><span class="i">$TotalSize</span><span class="s">)</span><span class="cm">,</span> <span class="q">&quot;\n&quot;</span><span class="sc">;</span>
185 172 <span class="s">}</span>
186 173 <span class="s">}</span>
187 174
188 175
189 176 <span class="c"># Perform frequency analysis of sequence lengths</span>
190 <a name="PerformLengthFrequencyAnalysis-"></a> 177 <span class="k">sub </span><span class="m">PerformLengthFrequencyAnalysis</span> <span class="s">{</span>
191 178 <span class="k">my</span><span class="s">(</span><span class="i">$SequenceDataRef</span><span class="cm">,</span> <span class="i">$SequenceLengthsRef</span><span class="s">)</span> = <span class="i">@_</span><span class="sc">;</span>
192 179 <span class="k">my</span> <span class="s">(</span><span class="i">$ID</span><span class="cm">,</span> <span class="i">$SeqLen</span><span class="cm">,</span> <span class="i">$Sequence</span><span class="cm">,</span> <span class="i">$SequenceLenBin</span><span class="cm">,</span> <span class="i">$LenBin</span><span class="cm">,</span> <span class="i">$SequenceLenCount</span><span class="cm">,</span> <span class="i">@SequenceLengths</span><span class="cm">,</span> <span class="i">%SequenceLenFrequency</span><span class="s">)</span><span class="sc">;</span>
193 180
194 181 <span class="i">@SequenceLengths</span> = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
195 182 <span class="i">%SequenceLenFrequency</span> = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
196 183 <span class="k">for</span> <span class="i">$ID</span> <span class="s">(</span><span class="i">@</span>{<span class="i">$SequenceDataRef</span>-&gt;{<span class="w">IDs</span>}}<span class="s">)</span> <span class="s">{</span>
197 184 <span class="i">$Sequence</span> = <span class="i">$SequenceDataRef</span>-&gt;{<span class="w">Sequence</span>}{<span class="i">$ID</span>}<span class="sc">;</span>
198 185 <span class="i">$SeqLen</span> = <span class="i">GetSequenceLength</span><span class="s">(</span><span class="i">$Sequence</span><span class="cm">,</span> <span class="i">$OptionsInfo</span>{<span class="w">IgnoreGaps</span>}<span class="s">)</span><span class="sc">;</span>
199 186 <span class="k">push</span> <span class="i">@SequenceLengths</span><span class="cm">,</span> <span class="i">$SeqLen</span><span class="sc">;</span>
200 187 <span class="s">}</span>
201 188 <span class="k">if</span> <span class="s">(</span><span class="i">@</span>{<span class="i">$OptionsInfo</span>{<span class="w">BinRange</span>}}<span class="s">)</span> <span class="s">{</span>
202 189 <span class="i">%SequenceLenFrequency</span> = <span class="i">Frequency</span><span class="s">(</span>\<span class="i">@SequenceLengths</span><span class="cm">,</span> \<span class="i">@</span>{<span class="i">$OptionsInfo</span>{<span class="w">BinRange</span>}}<span class="s">)</span><span class="sc">;</span>
203 190 <span class="s">}</span>
204 191 <span class="k">else</span> <span class="s">{</span>
205 192 <span class="i">%SequenceLenFrequency</span> = <span class="i">Frequency</span><span class="s">(</span>\<span class="i">@SequenceLengths</span><span class="cm">,</span> <span class="i">$OptionsInfo</span>{<span class="w">NumOfBins</span>}<span class="s">)</span><span class="sc">;</span>
206 193 <span class="s">}</span>
207 194 <span class="k">print</span> <span class="q">&quot;\nDistribution of sequence lengths (LengthBin =&gt; Count):\n&quot;</span><span class="sc">;</span>
208 195 <span class="k">for</span> <span class="i">$SequenceLenBin</span> <span class="s">(</span><span class="k">sort</span> <span class="s">{</span> <span class="i">$a</span> &lt;=&gt; <span class="i">$b</span><span class="s">}</span> <span class="k">keys</span> <span class="i">%SequenceLenFrequency</span><span class="s">)</span> <span class="s">{</span>
209 196 <span class="i">$SequenceLenCount</span> = <span class="i">$SequenceLenFrequency</span>{<span class="i">$SequenceLenBin</span>}<span class="sc">;</span>
210 197 <span class="i">$LenBin</span> = <span class="k">sprintf</span><span class="s">(</span><span class="q">&quot;%.1f&quot;</span><span class="cm">,</span> <span class="i">$SequenceLenBin</span><span class="s">)</span> + <span class="n">0</span><span class="sc">;</span>
211 198 <span class="k">print</span> <span class="q">&quot;$LenBin =&gt; $SequenceLenCount; &quot;</span><span class="sc">;</span>
212 199 <span class="s">}</span>
213 200 <span class="k">print</span> <span class="q">&quot;\n&quot;</span><span class="sc">;</span>
214 201 <span class="s">}</span>
215 202
216 203 <span class="c"># Retrieve information about sequence files...</span>
217 <a name="RetrieveSequenceFilesInfo-"></a> 204 <span class="k">sub </span><span class="m">RetrieveSequenceFilesInfo</span> <span class="s">{</span>
218 205 <span class="k">my</span><span class="s">(</span><span class="i">$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">$FileFormat</span><span class="cm">,</span> <span class="i">$ModifiedTimeString</span><span class="cm">,</span> <span class="i">$ModifiedDateString</span><span class="s">)</span><span class="sc">;</span>
219 206
220 207 <span class="i">%SequenceFilesInfo</span> = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
221 208 <span class="i">@</span>{<span class="i">$SequenceFilesInfo</span>{<span class="w">FileOkay</span>}} = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
222 209 <span class="i">@</span>{<span class="i">$SequenceFilesInfo</span>{<span class="w">FileFormat</span>}} = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
223 210 <span class="i">@</span>{<span class="i">$SequenceFilesInfo</span>{<span class="w">FileSize</span>}} = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
224 211 <span class="i">@</span>{<span class="i">$SequenceFilesInfo</span>{<span class="w">FileLastModified</span>}} = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
225 212
226 213 <span class="j">FILELIST:</span> <span class="k">for</span> <span class="i">$Index</span> <span class="s">(</span><span class="n">0</span> .. <span class="i">$#SequenceFilesList</span><span class="s">)</span> <span class="s">{</span>
227 214 <span class="i">$SequenceFile</span> = <span class="i">$SequenceFilesList</span>[<span class="i">$Index</span>]<span class="sc">;</span>
228 215
229 216 <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">&quot;$SequenceFile&quot;</span><span class="s">)</span> <span class="s">{</span>
230 217 <span class="k">warn</span> <span class="q">&quot;Warning: Ignoring file $SequenceFile: Couldn&#39;t open it: $! \n&quot;</span><span class="sc">;</span>
231 218 <span class="k">next</span> <span class="j">FILELIST</span><span class="sc">;</span>
232 219 <span class="s">}</span>
233 220 <span class="k">close</span> <span class="w">SEQUENCEFILE</span><span class="sc">;</span>
234 221
235 222 <span class="i">$SequenceFilesInfo</span>{<span class="w">FileOkay</span>}[<span class="i">$Index</span>] = <span class="n">0</span><span class="sc">;</span>
236 223 <span class="i">$SequenceFilesInfo</span>{<span class="w">FileFormat</span>}[<span class="i">$Index</span>] = <span class="q">&#39;NotSupported&#39;</span><span class="sc">;</span>
237 224 <span class="i">$SequenceFilesInfo</span>{<span class="w">FileSize</span>}[<span class="i">$Index</span>] = <span class="n">0</span><span class="sc">;</span>
238 225 <span class="i">$SequenceFilesInfo</span>{<span class="w">FileLastModified</span>}[<span class="i">$Index</span>] = <span class="q">&#39;&#39;</span><span class="sc">;</span>
239 226
240 227 <span class="s">(</span><span class="i">$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>
241 228 <span class="k">if</span> <span class="s">(</span>!<span class="i">$FileSupported</span><span class="s">)</span> <span class="s">{</span>
242 229 <span class="k">warn</span> <span class="q">&quot;Warning: Ignoring file $SequenceFile: Sequence file format is not supported.\n&quot;</span><span class="sc">;</span>
243 230 <span class="k">next</span> <span class="j">FILELIST</span><span class="sc">;</span>
244 231 <span class="s">}</span>
245 232
246 233 <span class="i">$SequenceFilesInfo</span>{<span class="w">FileOkay</span>}[<span class="i">$Index</span>] = <span class="n">1</span><span class="sc">;</span>
247 234 <span class="i">$SequenceFilesInfo</span>{<span class="w">FileFormat</span>}[<span class="i">$Index</span>] = <span class="i">$FileFormat</span><span class="sc">;</span>
248 235 <span class="i">$SequenceFilesInfo</span>{<span class="w">FileSize</span>}[<span class="i">$Index</span>] = <span class="i">FileSize</span><span class="s">(</span><span class="i">$SequenceFile</span><span class="s">)</span><span class="sc">;</span>
249 236
250 237 <span class="s">(</span><span class="i">$ModifiedTimeString</span><span class="cm">,</span> <span class="i">$ModifiedDateString</span><span class="s">)</span> = <span class="i">FormattedFileModificationTimeAndDate</span><span class="s">(</span><span class="i">$SequenceFile</span><span class="s">)</span><span class="sc">;</span>
251 238 <span class="i">$SequenceFilesInfo</span>{<span class="w">FileLastModified</span>}[<span class="i">$Index</span>] = <span class="q">&quot;$ModifiedTimeString; $ModifiedDateString&quot;</span><span class="sc">;</span>
252 239 <span class="s">}</span>
253 240 <span class="s">}</span>
254 241
255 242 <span class="c"># Process option values...</span>
256 <a name="ProcessOptions-"></a> 243 <span class="k">sub </span><span class="m">ProcessOptions</span> <span class="s">{</span>
257 244
258 245 <span class="i">$OptionsInfo</span>{<span class="w">All</span>} = <span class="k">defined</span> <span class="i">$Options</span>{<span class="w">all</span>} ? <span class="i">$Options</span>{<span class="w">all</span>} <span class="co">:</span> <span class="k">undef</span><span class="sc">;</span>
259 246
260 247 <span class="i">$OptionsInfo</span>{<span class="w">Count</span>} = <span class="k">defined</span> <span class="i">$Options</span>{<span class="w">count</span>} ? <span class="i">$Options</span>{<span class="w">count</span>} <span class="co">:</span> <span class="k">undef</span><span class="sc">;</span>
261 248 <span class="i">$OptionsInfo</span>{<span class="w">DetailLevel</span>} = <span class="i">$Options</span>{<span class="w">detail</span>}<span class="sc">;</span>
262 249 <span class="i">$OptionsInfo</span>{<span class="w">Frequency</span>} = <span class="k">defined</span> <span class="i">$Options</span>{<span class="w">frequency</span>} ? <span class="i">$Options</span>{<span class="w">frequency</span>} <span class="co">:</span> <span class="k">undef</span><span class="sc">;</span>
263 250 <span class="i">$OptionsInfo</span>{<span class="w">FrequencyBins</span>} = <span class="k">defined</span> <span class="i">$Options</span>{<span class="w">frequencybins</span>} ? <span class="i">$Options</span>{<span class="w">frequencybins</span>} <span class="co">:</span> <span class="k">undef</span><span class="sc">;</span>
264 251 <span class="i">$OptionsInfo</span>{<span class="w">IgnoreGaps</span>} = <span class="k">defined</span> <span class="i">$Options</span>{<span class="w">ignoregaps</span>} ? <span class="i">$Options</span>{<span class="w">ignoregaps</span>} <span class="co">:</span> <span class="k">undef</span><span class="sc">;</span>
265 252 <span class="i">$OptionsInfo</span>{<span class="w">Longest</span>} = <span class="k">defined</span> <span class="i">$Options</span>{<span class="w">longest</span>} ? <span class="i">$Options</span>{<span class="w">longest</span>} <span class="co">:</span> <span class="k">undef</span><span class="sc">;</span>
266 253 <span class="i">$OptionsInfo</span>{<span class="w">Shortest</span>} = <span class="k">defined</span> <span class="i">$Options</span>{<span class="w">shortest</span>} ? <span class="i">$Options</span>{<span class="w">shortest</span>} <span class="co">:</span> <span class="k">undef</span><span class="sc">;</span>
267 254 <span class="i">$OptionsInfo</span>{<span class="w">SequenceLengths</span>} = <span class="k">defined</span> <span class="i">$Options</span>{<span class="w">sequencelengths</span>} ? <span class="i">$Options</span>{<span class="w">sequencelengths</span>} <span class="co">:</span> <span class="k">undef</span><span class="sc">;</span>
268 255
269 256 <span class="i">$OptionsInfo</span>{<span class="w">FrequencyAnalysis</span>} = <span class="s">(</span><span class="i">$Options</span>{<span class="w">all</span>} || <span class="i">$Options</span>{<span class="w">frequency</span>}<span class="s">)</span> ? <span class="n">1</span> <span class="co">:</span> <span class="n">0</span><span class="sc">;</span>
270 257 <span class="i">$OptionsInfo</span>{<span class="w">ListLongestSequence</span>} = <span class="s">(</span><span class="i">$Options</span>{<span class="w">all</span>} || <span class="i">$Options</span>{<span class="w">longest</span>}<span class="s">)</span> ? <span class="n">1</span> <span class="co">:</span> <span class="n">0</span><span class="sc">;</span>
271 258 <span class="i">$OptionsInfo</span>{<span class="w">ListShortestSequence</span>} = <span class="s">(</span><span class="i">$Options</span>{<span class="w">all</span>} || <span class="i">$Options</span>{<span class="w">shortest</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 <span class="i">$OptionsInfo</span>{<span class="w">ListSequenceLengths</span>} = <span class="s">(</span><span class="i">$Options</span>{<span class="w">all</span>} || <span class="i">$Options</span>{<span class="w">sequencelengths</span>}<span class="s">)</span> ? <span class="n">1</span> <span class="co">:</span> <span class="n">0</span><span class="sc">;</span>
273 260 <span class="i">$OptionsInfo</span>{<span class="w">IgnoreGaps</span>} = <span class="s">(</span><span class="i">$Options</span>{<span class="w">ignoregaps</span>} =~ <span class="q">/Yes/i</span><span class="s">)</span> ? <span class="n">1</span> <span class="co">:</span> <span class="n">0</span><span class="sc">;</span>
274 261
275 262 <span class="c"># Setup frequency bin values...</span>
276 263 <span class="i">$OptionsInfo</span>{<span class="w">NumOfBins</span>} = <span class="n">4</span><span class="sc">;</span>
277 264 <span class="i">@</span>{<span class="i">$OptionsInfo</span>{<span class="w">BinRange</span>}} = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
278 265
279 266 <span class="k">if</span> <span class="s">(</span><span class="i">$Options</span>{<span class="w">frequencybins</span>} =~ <span class="q">/\,/</span><span class="s">)</span> <span class="s">{</span>
280 267 <span class="k">my</span><span class="s">(</span><span class="i">$BinValue</span><span class="cm">,</span> <span class="i">@SpecifiedBinRange</span><span class="s">)</span><span class="sc">;</span>
281 268 <span class="i">@SpecifiedBinRange</span> = <span class="k">split</span> <span class="q">/\,/</span><span class="cm">,</span> <span class="i">$Options</span>{<span class="w">frequencybins</span>}<span class="sc">;</span>
282 269 <span class="k">if</span> <span class="s">(</span><span class="i">@SpecifiedBinRange</span> &lt; <span class="n">2</span><span class="s">)</span> <span class="s">{</span>
283 270 <span class="k">die</span> <span class="q">&quot;Error: The value specified, $Options{frequencybins}, for option \&quot;--frequencybins\&quot; is not valid: Must contain at least two values. \n&quot;</span><span class="sc">;</span>
284 271 <span class="s">}</span>
285 272 <span class="k">for</span> <span class="i">$BinValue</span> <span class="s">(</span><span class="i">@SpecifiedBinRange</span><span class="s">)</span> <span class="s">{</span>
286 273 <span class="k">if</span> <span class="s">(</span>!<span class="i">IsNumerical</span><span class="s">(</span><span class="i">$BinValue</span><span class="s">)</span><span class="s">)</span> <span class="s">{</span>
287 274 <span class="k">die</span> <span class="q">&quot;Error: The value specified, $Options{frequencybins}, for option \&quot;--frequencybins\&quot; is not valid: Contains non numeric values. \n&quot;</span><span class="sc">;</span>
288 275 <span class="s">}</span>
289 276 <span class="s">}</span>
290 277 <span class="k">my</span><span class="s">(</span><span class="i">$Index1</span><span class="cm">,</span> <span class="i">$Index2</span><span class="s">)</span><span class="sc">;</span>
291 278 <span class="k">for</span> <span class="i">$Index1</span> <span class="s">(</span><span class="n">0</span> .. <span class="i">$#SpecifiedBinRange</span><span class="s">)</span> <span class="s">{</span>
292 279 <span class="k">for</span> <span class="i">$Index2</span> <span class="s">(</span><span class="s">(</span><span class="i">$Index1</span> + <span class="n">1</span><span class="s">)</span> .. <span class="i">$#SpecifiedBinRange</span><span class="s">)</span> <span class="s">{</span>
293 280 <span class="k">if</span> <span class="s">(</span><span class="i">$SpecifiedBinRange</span>[<span class="i">$Index1</span>] &gt;= <span class="i">$SpecifiedBinRange</span>[<span class="i">$Index2</span>]<span class="s">)</span> <span class="s">{</span>
294 281 <span class="k">die</span> <span class="q">&quot;Error: The value specified, $Options{frequencybins}, for option \&quot;--frequencybins\&quot; is not valid: Must contain values in ascending order. \n&quot;</span><span class="sc">;</span>
295 282 <span class="s">}</span>
296 283 <span class="s">}</span>
297 284 <span class="s">}</span>
298 285 <span class="k">push</span> <span class="i">@</span>{<span class="i">$OptionsInfo</span>{<span class="w">BinRange</span>}}<span class="cm">,</span> <span class="i">@SpecifiedBinRange</span><span class="sc">;</span>
299 286 <span class="s">}</span>
300 287 <span class="k">else</span> <span class="s">{</span>
301 288 <span class="i">$OptionsInfo</span>{<span class="w">NumOfBins</span>} = <span class="i">$Options</span>{<span class="w">frequencybins</span>}<span class="sc">;</span>
302 289 <span class="k">if</span> <span class="s">(</span>!<span class="i">IsPositiveInteger</span><span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">NumOfBins</span>}<span class="s">)</span><span class="s">)</span> <span class="s">{</span>
303 290 <span class="k">die</span> <span class="q">&quot;Error: The value specified, $Options{frequencybins}, for option \&quot;--frequencybins\&quot; is not valid. Allowed values: positive integer or \&quot;number,number,[number]...\&quot;. \n&quot;</span><span class="sc">;</span>
304 291 <span class="s">}</span>
305 292 <span class="s">}</span>
306 293 <span class="s">}</span>
307 294
308 295 <span class="c"># Setup script usage and retrieve command line arguments specified using various options...</span>
309 <a name="SetupScriptUsage-"></a> 296 <span class="k">sub </span><span class="m">SetupScriptUsage</span> <span class="s">{</span>
310 297
311 298 <span class="c"># Retrieve all the options...</span>
312 299 <span class="i">%Options</span> = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
313 300 <span class="i">$Options</span>{<span class="w">detail</span>} = <span class="n">1</span><span class="sc">;</span>
314 301 <span class="i">$Options</span>{<span class="w">ignoregaps</span>} = <span class="q">&#39;no&#39;</span><span class="sc">;</span>
315 302 <span class="i">$Options</span>{<span class="w">frequencybins</span>} = <span class="n">10</span><span class="sc">;</span>
316 303
317 304 <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">&quot;all|a&quot;</span><span class="cm">,</span> <span class="q">&quot;count|c&quot;</span><span class="cm">,</span> <span class="q">&quot;detail|d=i&quot;</span><span class="cm">,</span> <span class="q">&quot;frequency|f&quot;</span><span class="cm">,</span> <span class="q">&quot;frequencybins=s&quot;</span><span class="cm">,</span> <span class="q">&quot;help|h&quot;</span><span class="cm">,</span> <span class="q">&quot;ignoregaps|i=s&quot;</span><span class="cm">,</span> <span class="q">&quot;longest|l&quot;</span><span class="cm">,</span> <span class="q">&quot;shortest|s&quot;</span><span class="cm">,</span> <span class="q">&quot;sequencelengths&quot;</span><span class="cm">,</span> <span class="q">&quot;workingdir|w=s&quot;</span><span class="s">)</span><span class="s">)</span> <span class="s">{</span>
318 305 <span class="k">die</span> <span class="q">&quot;\nTo get a list of valid options and their values, use \&quot;$ScriptName -h\&quot; or\n\&quot;perl -S $ScriptName -h\&quot; command and try again...\n&quot;</span><span class="sc">;</span>
319 306 <span class="s">}</span>
320 307 <span class="k">if</span> <span class="s">(</span><span class="i">$Options</span>{<span class="w">workingdir</span>}<span class="s">)</span> <span class="s">{</span>
321 308 <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>
322 309 <span class="k">die</span> <span class="q">&quot;Error: The value specified, $Options{workingdir}, for option \&quot;-w --workingdir\&quot; is not a directory name.\n&quot;</span><span class="sc">;</span>
323 310 <span class="s">}</span>
324 311 <span class="k">chdir</span> <span class="i">$Options</span>{<span class="w">workingdir</span>} <span class="k">or</span> <span class="k">die</span> <span class="q">&quot;Error: Couldn&#39;t chdir $Options{workingdir}: $! \n&quot;</span><span class="sc">;</span>
325 312 <span class="s">}</span>
326 313 <span class="k">if</span> <span class="s">(</span>!<span class="i">IsPositiveInteger</span><span class="s">(</span><span class="i">$Options</span>{<span class="w">detail</span>}<span class="s">)</span><span class="s">)</span> <span class="s">{</span>
327 314 <span class="k">die</span> <span class="q">&quot;Error: The value specified, $Options{detail}, for option \&quot;-d --detail\&quot; is not valid. Allowed values: &gt; 0\n&quot;</span><span class="sc">;</span>
328 315 <span class="s">}</span>
329 316 <span class="k">if</span> <span class="s">(</span><span class="i">$Options</span>{<span class="w">ignoregaps</span>} !~ <span class="q">/^(yes|no)$/i</span><span class="s">)</span> <span class="s">{</span>
330 317 <span class="k">die</span> <span class="q">&quot;Error: The value specified, $Options{ignoregaps}, for option \&quot;-i --IgnoreGaps\&quot; is not valid. Allowed values: yes or no\n&quot;</span><span class="sc">;</span>
331 318 <span class="s">}</span>
332 319 <span class="s">}</span>
333 320
334 <a name="EOF-"></a></pre>
335 <p>&nbsp;</p>
336 <br />
337 <center>
338 <img src="../../../images/h2o2.png">
339 </center>
340 </body>
341 </html>