0
|
1 <html>
|
|
2 <head>
|
|
3 <title>MayaChemTools:Code:SequenceFileUtil.pm</title>
|
|
4 <meta http-equiv="content-type" content="text/html;charset=utf-8">
|
|
5 <link rel="stylesheet" type="text/css" href="../../../css/MayaChemToolsCode.css">
|
|
6 </head>
|
|
7 <body leftmargin="20" rightmargin="20" topmargin="10" bottommargin="10">
|
|
8 <br/>
|
|
9 <center>
|
|
10 <a href="http://www.mayachemtools.org" title="MayaChemTools Home"><img src="../../../images/MayaChemToolsLogo.gif" border="0" alt="MayaChemTools"></a>
|
|
11 </center>
|
|
12 <br/>
|
|
13 <pre>
|
|
14 <a name="package-SequenceFileUtil-"></a> 1 <span class="k">package </span><span class="i">SequenceFileUtil</span><span class="sc">;</span>
|
|
15 2 <span class="c">#</span>
|
|
16 3 <span class="c"># $RCSfile: SequenceFileUtil.pm,v $</span>
|
|
17 4 <span class="c"># $Date: 2015/02/28 20:47:18 $</span>
|
|
18 5 <span class="c"># $Revision: 1.33 $</span>
|
|
19 6 <span class="c">#</span>
|
|
20 7 <span class="c"># Author: Manish Sud <msud@san.rr.com></span>
|
|
21 8 <span class="c">#</span>
|
|
22 9 <span class="c"># Copyright (C) 2015 Manish Sud. All rights reserved.</span>
|
|
23 10 <span class="c">#</span>
|
|
24 11 <span class="c"># This file is part of MayaChemTools.</span>
|
|
25 12 <span class="c">#</span>
|
|
26 13 <span class="c"># MayaChemTools is free software; you can redistribute it and/or modify it under</span>
|
|
27 14 <span class="c"># the terms of the GNU Lesser General Public License as published by the Free</span>
|
|
28 15 <span class="c"># Software Foundation; either version 3 of the License, or (at your option) any</span>
|
|
29 16 <span class="c"># later version.</span>
|
|
30 17 <span class="c">#</span>
|
|
31 18 <span class="c"># MayaChemTools is distributed in the hope that it will be useful, but without</span>
|
|
32 19 <span class="c"># any warranty; without even the implied warranty of merchantability of fitness</span>
|
|
33 20 <span class="c"># for a particular purpose. See the GNU Lesser General Public License for more</span>
|
|
34 21 <span class="c"># details.</span>
|
|
35 22 <span class="c">#</span>
|
|
36 23 <span class="c"># You should have received a copy of the GNU Lesser General Public License</span>
|
|
37 24 <span class="c"># along with MayaChemTools; if not, see <http://www.gnu.org/licenses/> or</span>
|
|
38 25 <span class="c"># write to the Free Software Foundation Inc., 59 Temple Place, Suite 330,</span>
|
|
39 26 <span class="c"># Boston, MA, 02111-1307, USA.</span>
|
|
40 27 <span class="c">#</span>
|
|
41 28
|
|
42 29 <span class="k">use</span> <span class="w">strict</span><span class="sc">;</span>
|
|
43 30 <span class="k">use</span> <span class="w">Exporter</span><span class="sc">;</span>
|
|
44 31 <span class="k">use</span> <span class="w">Text::ParseWords</span><span class="sc">;</span>
|
|
45 32 <span class="k">use</span> <span class="w">TextUtil</span><span class="sc">;</span>
|
|
46 33 <span class="k">use</span> <span class="w">FileUtil</span><span class="sc">;</span>
|
|
47 34
|
|
48 35 <span class="k">use</span> <span class="w">vars</span> <span class="q">qw(@ISA @EXPORT @EXPORT_OK %EXPORT_TAGS)</span><span class="sc">;</span>
|
|
49 36
|
|
50 37 <span class="i">@ISA</span> = <span class="q">qw(Exporter)</span><span class="sc">;</span>
|
|
51 38 <span class="i">@EXPORT</span> = <span class="q">qw(AreSequenceLengthsIdentical CalcuatePercentSequenceIdentity CalculatePercentSequenceIdentityMatrix GetLongestSequence GetShortestSequence GetSequenceLength IsGapResidue IsSupportedSequenceFile IsClustalWSequenceFile IsPearsonFastaSequenceFile IsMSFSequenceFile ReadSequenceFile RemoveSequenceGaps RemoveSequenceAlignmentGapColumns WritePearsonFastaSequenceFile)</span><span class="sc">;</span>
|
|
52 39 <span class="i">@EXPORT_OK</span> = <span class="q">qw()</span><span class="sc">;</span>
|
|
53 40
|
|
54 41 <span class="i">%EXPORT_TAGS</span> = <span class="s">(</span><span class="w">all</span> <span class="cm">=></span> <span class="s">[</span><span class="i">@EXPORT</span><span class="cm">,</span> <span class="i">@EXPORT_OK</span><span class="s">]</span><span class="s">)</span><span class="sc">;</span>
|
|
55 42
|
|
56 43 <span class="c"># Compare lengths of all sequences...</span>
|
|
57 <a name="AreSequenceLengthsIdentical-"></a> 44 <span class="k">sub </span><span class="m">AreSequenceLengthsIdentical</span> <span class="s">{</span>
|
|
58 45 <span class="k">my</span><span class="s">(</span><span class="i">$SequencesDataRef</span><span class="s">)</span> = <span class="i">@_</span><span class="sc">;</span>
|
|
59 46 <span class="k">my</span><span class="s">(</span><span class="i">$Status</span><span class="cm">,</span> <span class="i">$ID</span><span class="cm">,</span> <span class="i">$FirstID</span><span class="cm">,</span> <span class="i">$FirstSeqLen</span><span class="cm">,</span> <span class="i">$FirstDifferentLenID</span><span class="cm">,</span> <span class="i">$SeqLen</span><span class="s">)</span><span class="sc">;</span>
|
|
60 47
|
|
61 48 <span class="i">$Status</span> = <span class="n">1</span><span class="sc">;</span>
|
|
62 49 <span class="i">$FirstID</span> = <span class="q">''</span><span class="sc">;</span>
|
|
63 50 <span class="i">$FirstDifferentLenID</span> = <span class="q">''</span><span class="sc">;</span>
|
|
64 51
|
|
65 52 <span class="j">ID:</span> <span class="k">for</span> <span class="i">$ID</span> <span class="s">(</span><span class="i">@</span>{<span class="i">$SequencesDataRef</span>->{<span class="w">IDs</span>}}<span class="s">)</span> <span class="s">{</span>
|
|
66 53 <span class="k">if</span> <span class="s">(</span>!<span class="i">$FirstID</span><span class="s">)</span> <span class="s">{</span>
|
|
67 54 <span class="i">$FirstID</span> = <span class="i">$ID</span><span class="sc">;</span>
|
|
68 55 <span class="i">$FirstSeqLen</span> = <span class="k">length</span><span class="s">(</span><span class="i">$SequencesDataRef</span>->{<span class="w">Sequence</span>}{<span class="i">$ID</span>}<span class="s">)</span><span class="sc">;</span>
|
|
69 56 <span class="k">next</span> <span class="j">ID</span><span class="sc">;</span>
|
|
70 57 <span class="s">}</span>
|
|
71 58 <span class="i">$SeqLen</span> = <span class="k">length</span><span class="s">(</span><span class="i">$SequencesDataRef</span>->{<span class="w">Sequence</span>}{<span class="i">$ID</span>}<span class="s">)</span><span class="sc">;</span>
|
|
72 59 <span class="k">if</span> <span class="s">(</span><span class="i">$SeqLen</span> != <span class="i">$FirstSeqLen</span><span class="s">)</span> <span class="s">{</span>
|
|
73 60 <span class="i">$Status</span> = <span class="n">0</span><span class="sc">;</span>
|
|
74 61 <span class="i">$FirstDifferentLenID</span> = <span class="i">$ID</span><span class="sc">;</span>
|
|
75 62 <span class="k">last</span> <span class="j">ID</span><span class="sc">;</span>
|
|
76 63 <span class="s">}</span>
|
|
77 64 <span class="s">}</span>
|
|
78 65 <span class="k">return</span> <span class="s">(</span><span class="i">$Status</span><span class="s">)</span><span class="sc">;</span>
|
|
79 66 <span class="s">}</span>
|
|
80 67
|
|
81 68 <span class="c"># Calculate percent identity between two sequences. By default, gaps are ignored.</span>
|
|
82 <a name="CalcuatePercentSequenceIdentity-"></a> 69 <span class="k">sub </span><span class="m">CalcuatePercentSequenceIdentity</span> <span class="s">{</span>
|
|
83 70 <span class="k">my</span><span class="s">(</span><span class="i">$Sequence1</span><span class="cm">,</span> <span class="i">$Sequence2</span><span class="cm">,</span> <span class="i">$PercentIdentity</span><span class="cm">,</span> <span class="i">$IgnoreGaps</span><span class="cm">,</span> <span class="i">$Precision</span><span class="s">)</span><span class="sc">;</span>
|
|
84 71
|
|
85 72 <span class="i">$PercentIdentity</span> = <span class="q">''</span><span class="sc">;</span>
|
|
86 73 <span class="i">$Precision</span> = <span class="n">1</span><span class="sc">;</span>
|
|
87 74 <span class="i">$IgnoreGaps</span> = <span class="n">1</span><span class="sc">;</span>
|
|
88 75 <span class="k">if</span> <span class="s">(</span><span class="i">@_</span> == <span class="n">4</span><span class="s">)</span> <span class="s">{</span>
|
|
89 76 <span class="s">(</span><span class="i">$Sequence1</span><span class="cm">,</span> <span class="i">$Sequence2</span><span class="cm">,</span> <span class="i">$IgnoreGaps</span><span class="cm">,</span> <span class="i">$Precision</span><span class="s">)</span> = <span class="i">@_</span><span class="sc">;</span>
|
|
90 77 <span class="s">}</span>
|
|
91 78 <span class="k">elsif</span> <span class="s">(</span><span class="i">@_</span> == <span class="n">3</span><span class="s">)</span> <span class="s">{</span>
|
|
92 79 <span class="s">(</span><span class="i">$Sequence1</span><span class="cm">,</span> <span class="i">$Sequence2</span><span class="cm">,</span> <span class="i">$IgnoreGaps</span><span class="s">)</span> = <span class="i">@_</span><span class="sc">;</span>
|
|
93 80 <span class="s">}</span>
|
|
94 81 <span class="k">elsif</span> <span class="s">(</span><span class="i">@_</span> == <span class="n">2</span><span class="s">)</span> <span class="s">{</span>
|
|
95 82 <span class="s">(</span><span class="i">$Sequence1</span><span class="cm">,</span> <span class="i">$Sequence2</span><span class="s">)</span> = <span class="i">@_</span><span class="sc">;</span>
|
|
96 83 <span class="s">}</span>
|
|
97 84 <span class="k">else</span> <span class="s">{</span>
|
|
98 85 <span class="k">return</span> <span class="i">$PercentIdentity</span><span class="sc">;</span>
|
|
99 86 <span class="s">}</span>
|
|
100 87 <span class="k">if</span> <span class="s">(</span>!<span class="s">(</span><span class="i">IsNotEmpty</span><span class="s">(</span><span class="i">$Sequence1</span><span class="s">)</span> && <span class="i">IsNotEmpty</span><span class="s">(</span><span class="i">$Sequence2</span><span class="s">)</span><span class="s">)</span><span class="s">)</span> <span class="s">{</span>
|
|
101 88 <span class="k">return</span> <span class="i">$PercentIdentity</span><span class="sc">;</span>
|
|
102 89 <span class="s">}</span>
|
|
103 90 <span class="k">my</span><span class="s">(</span><span class="i">$Index</span><span class="cm">,</span> <span class="i">$Identity</span><span class="cm">,</span> <span class="i">$Sequence1Len</span><span class="cm">,</span> <span class="i">$Sequence2Len</span><span class="cm">,</span> <span class="i">$Residue1</span><span class="cm">,</span> <span class="i">$Residue2</span><span class="cm">,</span> <span class="i">$ResMatchCount</span><span class="cm">,</span> <span class="i">$ResCount</span><span class="s">)</span><span class="sc">;</span>
|
|
104 91
|
|
105 92 <span class="i">$Sequence1Len</span> = <span class="k">length</span><span class="s">(</span><span class="i">$Sequence1</span><span class="s">)</span><span class="sc">;</span>
|
|
106 93 <span class="i">$Sequence2Len</span> = <span class="k">length</span><span class="s">(</span><span class="i">$Sequence2</span><span class="s">)</span><span class="sc">;</span>
|
|
107 94
|
|
108 95 <span class="i">$ResMatchCount</span> = <span class="n">0</span><span class="sc">;</span>
|
|
109 96 <span class="i">$ResCount</span> = <span class="n">0</span><span class="sc">;</span>
|
|
110 97 <span class="j">RESIDUE:</span> <span class="k">for</span> <span class="i">$Index</span> <span class="s">(</span><span class="n">0</span> .. <span class="s">(</span><span class="i">$Sequence1Len</span> - <span class="n">1</span><span class="s">)</span><span class="s">)</span> <span class="s">{</span>
|
|
111 98 <span class="i">$Residue1</span> = <span class="k">substr</span><span class="s">(</span><span class="i">$Sequence1</span><span class="cm">,</span> <span class="i">$Index</span><span class="cm">,</span> <span class="n">1</span><span class="s">)</span><span class="sc">;</span>
|
|
112 99 <span class="i">$Residue2</span> = <span class="s">(</span><span class="i">$Index</span> < <span class="i">$Sequence2Len</span><span class="s">)</span> ? <span class="k">substr</span><span class="s">(</span><span class="i">$Sequence2</span><span class="cm">,</span> <span class="i">$Index</span><span class="cm">,</span> <span class="n">1</span><span class="s">)</span> <span class="co">:</span> <span class="q">''</span><span class="sc">;</span>
|
|
113 100 <span class="k">if</span> <span class="s">(</span><span class="i">$IgnoreGaps</span><span class="s">)</span> <span class="s">{</span>
|
|
114 101 <span class="k">if</span> <span class="s">(</span><span class="i">$Residue1</span> !~ <span class="q">/[A-Z]/i</span> || <span class="i">$Residue2</span> !~ <span class="q">/[A-Z]/i</span><span class="s">)</span> <span class="s">{</span>
|
|
115 102 <span class="k">next</span> <span class="j">RESIDUE</span><span class="sc">;</span>
|
|
116 103 <span class="s">}</span>
|
|
117 104 <span class="s">}</span>
|
|
118 105 <span class="k">if</span> <span class="s">(</span><span class="i">$Residue1</span> <span class="k">eq</span> <span class="i">$Residue2</span><span class="s">)</span> <span class="s">{</span>
|
|
119 106 <span class="i">$ResMatchCount</span>++<span class="sc">;</span>
|
|
120 107 <span class="s">}</span>
|
|
121 108 <span class="i">$ResCount</span>++<span class="sc">;</span>
|
|
122 109 <span class="s">}</span>
|
|
123 110 <span class="i">$Identity</span> = <span class="i">$ResCount</span> ? <span class="s">(</span><span class="i">$ResMatchCount</span>/<span class="i">$ResCount</span><span class="s">)</span> <span class="co">:</span> <span class="n">0.0</span><span class="sc">;</span>
|
|
124 111 <span class="i">$PercentIdentity</span> = <span class="k">sprintf</span><span class="s">(</span><span class="q">"%.${Precision}f"</span><span class="cm">,</span> <span class="s">(</span><span class="i">$Identity</span> * <span class="n">100</span><span class="s">)</span><span class="s">)</span><span class="sc">;</span>
|
|
125 112
|
|
126 113 <span class="k">return</span> <span class="i">$PercentIdentity</span><span class="sc">;</span>
|
|
127 114 <span class="s">}</span>
|
|
128 115
|
|
129 116 <span class="c"># Calculate pairwise identify matrix for all the sequences and return a reference</span>
|
|
130 117 <span class="c"># to a hash with the following keys:</span>
|
|
131 118 <span class="c">#</span>
|
|
132 119 <span class="c"># {IDs} - Sequence IDs</span>
|
|
133 120 <span class="c"># {Count} - Number of IDs</span>
|
|
134 121 <span class="c"># {PercentIdentity}{$RowID}{$ColID} - Percent identify for a pair of sequences</span>
|
|
135 122 <span class="c">#</span>
|
|
136 <a name="CalculatePercentSequenceIdentityMatrix-"></a> 123 <span class="k">sub </span><span class="m">CalculatePercentSequenceIdentityMatrix</span> <span class="s">{</span>
|
|
137 124 <span class="k">my</span><span class="s">(</span><span class="i">$SequencesDataRef</span><span class="cm">,</span> <span class="i">$IgnoreGaps</span><span class="cm">,</span> <span class="cm">,</span> <span class="i">$Precision</span><span class="cm">,</span> <span class="i">$ID</span><span class="cm">,</span> <span class="i">$RowID</span><span class="cm">,</span> <span class="i">$ColID</span><span class="cm">,</span> <span class="i">$RowIDSeq</span><span class="cm">,</span> <span class="i">$ColIDSeq</span><span class="cm">,</span> <span class="i">$PercentIdentity</span><span class="cm">,</span> <span class="i">%IdentityMatrixData</span><span class="s">)</span><span class="sc">;</span>
|
|
138 125
|
|
139 126 <span class="i">$IgnoreGaps</span> = <span class="n">1</span><span class="sc">;</span>
|
|
140 127 <span class="i">$Precision</span> = <span class="n">1</span><span class="sc">;</span>
|
|
141 128 <span class="k">if</span> <span class="s">(</span><span class="i">@_</span> == <span class="n">3</span><span class="s">)</span> <span class="s">{</span>
|
|
142 129 <span class="s">(</span><span class="i">$SequencesDataRef</span><span class="cm">,</span> <span class="i">$IgnoreGaps</span><span class="cm">,</span> <span class="i">$Precision</span><span class="s">)</span> = <span class="i">@_</span><span class="sc">;</span>
|
|
143 130 <span class="s">}</span>
|
|
144 131 <span class="k">elsif</span> <span class="s">(</span><span class="i">@_</span> == <span class="n">2</span><span class="s">)</span> <span class="s">{</span>
|
|
145 132 <span class="s">(</span><span class="i">$SequencesDataRef</span><span class="cm">,</span> <span class="i">$IgnoreGaps</span><span class="s">)</span> = <span class="i">@_</span><span class="sc">;</span>
|
|
146 133 <span class="s">}</span>
|
|
147 134 <span class="k">else</span> <span class="s">{</span>
|
|
148 135 <span class="s">(</span><span class="i">$SequencesDataRef</span><span class="s">)</span> = <span class="i">@_</span><span class="sc">;</span>
|
|
149 136 <span class="s">}</span>
|
|
150 137
|
|
151 138 <span class="i">%IdentityMatrixData</span> = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
152 139 <span class="i">@</span>{<span class="i">$IdentityMatrixData</span>{<span class="w">IDs</span>}} = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
153 140 <span class="i">%</span>{<span class="i">$IdentityMatrixData</span>{<span class="w">PercentIdentity</span>}} = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
154 141 <span class="i">$IdentityMatrixData</span>{<span class="w">Count</span>} = <span class="n">0</span><span class="sc">;</span>
|
|
155 142
|
|
156 143 <span class="k">for</span> <span class="i">$ID</span> <span class="s">(</span><span class="i">@</span>{<span class="i">$SequencesDataRef</span>->{<span class="w">IDs</span>}}<span class="s">)</span> <span class="s">{</span>
|
|
157 144 <span class="k">push</span> <span class="i">@</span>{<span class="i">$IdentityMatrixData</span>{<span class="w">IDs</span>}}<span class="cm">,</span> <span class="i">$ID</span><span class="sc">;</span>
|
|
158 145 <span class="i">$IdentityMatrixData</span>{<span class="w">Count</span>} += <span class="n">1</span><span class="sc">;</span>
|
|
159 146 <span class="s">}</span>
|
|
160 147 <span class="c"># Initialize and calculate percent identity data values...</span>
|
|
161 148 <span class="k">for</span> <span class="i">$RowID</span> <span class="s">(</span><span class="i">@</span>{<span class="i">$SequencesDataRef</span>->{<span class="w">IDs</span>}}<span class="s">)</span> <span class="s">{</span>
|
|
162 149 <span class="i">%</span>{<span class="i">$IdentityMatrixData</span>{<span class="w">PercentIdentity</span>}{<span class="i">$RowID</span>}} = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
163 150 <span class="i">$RowIDSeq</span> = <span class="i">$SequencesDataRef</span>->{<span class="w">Sequence</span>}{<span class="i">$RowID</span>}<span class="sc">;</span>
|
|
164 151 <span class="k">for</span> <span class="i">$ColID</span> <span class="s">(</span><span class="i">@</span>{<span class="i">$SequencesDataRef</span>->{<span class="w">IDs</span>}}<span class="s">)</span> <span class="s">{</span>
|
|
165 152 <span class="i">$IdentityMatrixData</span>{<span class="i">$RowID</span>}{<span class="i">$ColID</span>} = <span class="q">''</span><span class="sc">;</span>
|
|
166 153 <span class="i">$ColIDSeq</span> = <span class="i">$SequencesDataRef</span>->{<span class="w">Sequence</span>}{<span class="i">$ColID</span>}<span class="sc">;</span>
|
|
167 154 <span class="i">$PercentIdentity</span> = <span class="i">CalcuatePercentSequenceIdentity</span><span class="s">(</span><span class="i">$RowIDSeq</span><span class="cm">,</span> <span class="i">$ColIDSeq</span><span class="cm">,</span> <span class="i">$IgnoreGaps</span><span class="cm">,</span> <span class="i">$Precision</span><span class="s">)</span><span class="sc">;</span>
|
|
168 155 <span class="i">$IdentityMatrixData</span>{<span class="w">PercentIdentity</span>}{<span class="i">$RowID</span>}{<span class="i">$ColID</span>} = <span class="i">$PercentIdentity</span><span class="sc">;</span>
|
|
169 156 <span class="s">}</span>
|
|
170 157 <span class="s">}</span>
|
|
171 158 <span class="k">return</span> \<span class="i">%IdentityMatrixData</span><span class="sc">;</span>
|
|
172 159 <span class="s">}</span>
|
|
173 160
|
|
174 161 <span class="c"># Retrieve information about shortest sequence...</span>
|
|
175 <a name="GetShortestSequence-"></a> 162 <span class="k">sub </span><span class="m">GetShortestSequence</span> <span class="s">{</span>
|
|
176 163 <span class="k">my</span><span class="s">(</span><span class="i">$SequencesDataRef</span><span class="cm">,</span> <span class="i">$IgnoreGaps</span><span class="cm">,</span> <span class="i">$ID</span><span class="cm">,</span> <span class="i">$Sequence</span><span class="cm">,</span> <span class="i">$SeqLen</span><span class="cm">,</span> <span class="i">$Description</span><span class="s">)</span><span class="sc">;</span>
|
|
177 164
|
|
178 165 <span class="i">$IgnoreGaps</span> = <span class="n">1</span><span class="sc">;</span>
|
|
179 166 <span class="k">if</span> <span class="s">(</span><span class="i">@_</span> == <span class="n">2</span><span class="s">)</span> <span class="s">{</span>
|
|
180 167 <span class="s">(</span><span class="i">$SequencesDataRef</span><span class="cm">,</span> <span class="i">$IgnoreGaps</span><span class="s">)</span> = <span class="i">@_</span><span class="sc">;</span>
|
|
181 168 <span class="s">}</span>
|
|
182 169 <span class="k">else</span> <span class="s">{</span>
|
|
183 170 <span class="s">(</span><span class="i">$SequencesDataRef</span><span class="s">)</span> = <span class="i">@_</span><span class="sc">;</span>
|
|
184 171 <span class="s">}</span>
|
|
185 172
|
|
186 173 <span class="s">(</span><span class="i">$ID</span><span class="cm">,</span> <span class="i">$Sequence</span><span class="cm">,</span> <span class="i">$SeqLen</span><span class="cm">,</span> <span class="i">$Description</span><span class="s">)</span> = <span class="i">_GetShortestOrLongestSequence</span><span class="s">(</span><span class="i">$SequencesDataRef</span><span class="cm">,</span> <span class="q">'Shortest'</span><span class="cm">,</span> <span class="i">$IgnoreGaps</span><span class="s">)</span><span class="sc">;</span>
|
|
187 174 <span class="k">return</span> <span class="s">(</span><span class="i">$ID</span><span class="cm">,</span> <span class="i">$Sequence</span><span class="cm">,</span> <span class="i">$SeqLen</span><span class="cm">,</span> <span class="i">$Description</span><span class="s">)</span><span class="sc">;</span>
|
|
188 175 <span class="s">}</span>
|
|
189 176
|
|
190 177 <span class="c"># Retrieve information about longest sequence..</span>
|
|
191 <a name="GetLongestSequence-"></a> 178 <span class="k">sub </span><span class="m">GetLongestSequence</span> <span class="s">{</span>
|
|
192 179 <span class="k">my</span><span class="s">(</span><span class="i">$SequencesDataRef</span><span class="cm">,</span> <span class="i">$IgnoreGaps</span><span class="cm">,</span> <span class="i">$ID</span><span class="cm">,</span> <span class="i">$Sequence</span><span class="cm">,</span> <span class="i">$SeqLen</span><span class="cm">,</span> <span class="i">$Description</span><span class="s">)</span><span class="sc">;</span>
|
|
193 180
|
|
194 181 <span class="i">$IgnoreGaps</span> = <span class="n">1</span><span class="sc">;</span>
|
|
195 182 <span class="k">if</span> <span class="s">(</span><span class="i">@_</span> == <span class="n">2</span><span class="s">)</span> <span class="s">{</span>
|
|
196 183 <span class="s">(</span><span class="i">$SequencesDataRef</span><span class="cm">,</span> <span class="i">$IgnoreGaps</span><span class="s">)</span> = <span class="i">@_</span><span class="sc">;</span>
|
|
197 184 <span class="s">}</span>
|
|
198 185 <span class="k">else</span> <span class="s">{</span>
|
|
199 186 <span class="s">(</span><span class="i">$SequencesDataRef</span><span class="s">)</span> = <span class="i">@_</span><span class="sc">;</span>
|
|
200 187 <span class="s">}</span>
|
|
201 188
|
|
202 189 <span class="s">(</span><span class="i">$ID</span><span class="cm">,</span> <span class="i">$Sequence</span><span class="cm">,</span> <span class="i">$SeqLen</span><span class="cm">,</span> <span class="i">$Description</span><span class="s">)</span> = <span class="i">_GetShortestOrLongestSequence</span><span class="s">(</span><span class="i">$SequencesDataRef</span><span class="cm">,</span> <span class="q">'Longest'</span><span class="cm">,</span> <span class="i">$IgnoreGaps</span><span class="s">)</span><span class="sc">;</span>
|
|
203 190 <span class="k">return</span> <span class="s">(</span><span class="i">$ID</span><span class="cm">,</span> <span class="i">$Sequence</span><span class="cm">,</span> <span class="i">$SeqLen</span><span class="cm">,</span> <span class="i">$Description</span><span class="s">)</span><span class="sc">;</span>
|
|
204 191 <span class="s">}</span>
|
|
205 192
|
|
206 193 <span class="c"># Get sequence length...</span>
|
|
207 <a name="GetSequenceLength-"></a> 194 <span class="k">sub </span><span class="m">GetSequenceLength</span> <span class="s">{</span>
|
|
208 195 <span class="k">my</span><span class="s">(</span><span class="i">$Seq</span><span class="cm">,</span> <span class="i">$SeqLen</span><span class="cm">,</span> <span class="i">$IgnoreGaps</span><span class="s">)</span><span class="sc">;</span>
|
|
209 196
|
|
210 197 <span class="i">$SeqLen</span> = <span class="q">''</span><span class="sc">;</span> <span class="i">$IgnoreGaps</span> = <span class="n">1</span><span class="sc">;</span>
|
|
211 198 <span class="k">if</span> <span class="s">(</span><span class="i">@_</span> == <span class="n">2</span><span class="s">)</span> <span class="s">{</span>
|
|
212 199 <span class="s">(</span><span class="i">$Seq</span><span class="cm">,</span> <span class="i">$IgnoreGaps</span><span class="s">)</span> = <span class="i">@_</span><span class="sc">;</span>
|
|
213 200 <span class="s">}</span>
|
|
214 201 <span class="k">else</span> <span class="s">{</span>
|
|
215 202 <span class="s">(</span><span class="i">$Seq</span><span class="s">)</span> = <span class="i">@_</span><span class="sc">;</span>
|
|
216 203 <span class="s">}</span>
|
|
217 204 <span class="k">if</span> <span class="s">(</span><span class="i">$IgnoreGaps</span><span class="s">)</span> <span class="s">{</span>
|
|
218 205 <span class="k">my</span><span class="s">(</span><span class="i">$Index</span><span class="cm">,</span> <span class="i">$Residue</span><span class="s">)</span><span class="sc">;</span>
|
|
219 206 <span class="i">$SeqLen</span> = <span class="n">0</span><span class="sc">;</span>
|
|
220 207 <span class="k">for</span> <span class="i">$Index</span> <span class="s">(</span><span class="n">0</span> .. <span class="s">(</span><span class="k">length</span><span class="s">(</span><span class="i">$Seq</span><span class="s">)</span> - <span class="n">1</span><span class="s">)</span><span class="s">)</span> <span class="s">{</span>
|
|
221 208 <span class="i">$Residue</span> = <span class="k">substr</span><span class="s">(</span><span class="i">$Seq</span><span class="cm">,</span> <span class="i">$Index</span><span class="cm">,</span> <span class="n">1</span><span class="s">)</span><span class="sc">;</span>
|
|
222 209 <span class="k">if</span> <span class="s">(</span><span class="i">$Residue</span> =~ <span class="q">/[A-Z]/i</span><span class="s">)</span> <span class="s">{</span>
|
|
223 210 <span class="i">$SeqLen</span>++<span class="sc">;</span>
|
|
224 211 <span class="s">}</span>
|
|
225 212 <span class="s">}</span>
|
|
226 213 <span class="s">}</span>
|
|
227 214 <span class="k">else</span> <span class="s">{</span>
|
|
228 215 <span class="i">$SeqLen</span> = <span class="k">length</span><span class="s">(</span><span class="i">$Seq</span><span class="s">)</span><span class="sc">;</span>
|
|
229 216 <span class="s">}</span>
|
|
230 217
|
|
231 218 <span class="k">return</span> <span class="i">$SeqLen</span><span class="sc">;</span>
|
|
232 219 <span class="s">}</span>
|
|
233 220
|
|
234 221 <span class="c"># Is it a gap residue...</span>
|
|
235 <a name="IsGapResidue-"></a> 222 <span class="k">sub </span><span class="m">IsGapResidue</span> <span class="s">{</span>
|
|
236 223 <span class="k">my</span><span class="s">(</span><span class="i">$Residue</span><span class="s">)</span> = <span class="i">@_</span><span class="sc">;</span>
|
|
237 224 <span class="k">my</span><span class="s">(</span><span class="i">$Status</span><span class="s">)</span><span class="sc">;</span>
|
|
238 225
|
|
239 226 <span class="i">$Status</span> = <span class="s">(</span><span class="i">$Residue</span> !~ <span class="q">/[A-Z]/i</span> <span class="s">)</span> ? <span class="n">1</span> <span class="co">:</span> <span class="n">0</span><span class="sc">;</span>
|
|
240 227
|
|
241 228 <span class="k">return</span> <span class="i">$Status</span><span class="sc">;</span>
|
|
242 229 <span class="s">}</span>
|
|
243 230
|
|
244 231 <span class="c"># Is it a supported sequence file?</span>
|
|
245 232 <span class="c">#</span>
|
|
246 233 <span class="c"># Supported seqence formats are:</span>
|
|
247 234 <span class="c">#</span>
|
|
248 235 <span class="c"># ALN/ClustalW .aln</span>
|
|
249 236 <span class="c"># GCG/MSF .msf</span>
|
|
250 237 <span class="c"># PILEUP/MSF .msf</span>
|
|
251 238 <span class="c"># Fasts(Pearson) .fasta, .fta</span>
|
|
252 239 <span class="c"># NBRF/PIR .pir</span>
|
|
253 240 <span class="c">#</span>
|
|
254 <a name="IsSupportedSequenceFile-"></a> 241 <span class="k">sub </span><span class="m">IsSupportedSequenceFile</span> <span class="s">{</span>
|
|
255 242 <span class="k">my</span><span class="s">(</span><span class="i">$SequenceFile</span><span class="s">)</span> = <span class="i">@_</span><span class="sc">;</span>
|
|
256 243 <span class="k">my</span><span class="s">(</span><span class="i">$Status</span><span class="cm">,</span> <span class="i">$SequenceFormat</span><span class="s">)</span><span class="sc">;</span>
|
|
257 244 <span class="i">$Status</span> = <span class="n">0</span><span class="sc">;</span> <span class="i">$SequenceFormat</span> = <span class="q">'NotSupported'</span><span class="sc">;</span>
|
|
258 245
|
|
259 246 <span class="j">SEQFORMAT:</span> <span class="s">{</span>
|
|
260 247 <span class="k">if</span> <span class="s">(</span><span class="i">IsClustalWSequenceFile</span><span class="s">(</span><span class="i">$SequenceFile</span><span class="s">)</span><span class="s">)</span> <span class="s">{</span><span class="i">$Status</span> = <span class="n">1</span><span class="sc">;</span> <span class="i">$SequenceFormat</span> = <span class="q">'ClustalW'</span><span class="sc">;</span> <span class="k">last</span> <span class="j">SEQFORMAT</span><span class="s">}</span>
|
|
261 248 <span class="k">if</span> <span class="s">(</span><span class="i">IsPearsonFastaSequenceFile</span><span class="s">(</span><span class="i">$SequenceFile</span><span class="s">)</span><span class="s">)</span> <span class="s">{</span><span class="i">$Status</span> = <span class="n">1</span><span class="sc">;</span> <span class="i">$SequenceFormat</span> = <span class="q">'Pearson'</span><span class="sc">;</span> <span class="k">last</span> <span class="j">SEQFORMAT</span><span class="s">}</span>
|
|
262 249 <span class="k">if</span> <span class="s">(</span><span class="i">IsPIRFastaSequenceFile</span><span class="s">(</span><span class="i">$SequenceFile</span><span class="s">)</span><span class="s">)</span> <span class="s">{</span><span class="i">$Status</span> = <span class="n">1</span><span class="sc">;</span> <span class="i">$SequenceFormat</span> = <span class="q">'PIR'</span><span class="sc">;</span> <span class="k">last</span> <span class="j">SEQFORMAT</span><span class="s">}</span>
|
|
263 250 <span class="k">if</span> <span class="s">(</span><span class="i">IsMSFSequenceFile</span><span class="s">(</span><span class="i">$SequenceFile</span><span class="s">)</span><span class="s">)</span> <span class="s">{</span><span class="i">$Status</span> = <span class="n">1</span><span class="sc">;</span> <span class="i">$SequenceFormat</span> = <span class="q">'MSF'</span><span class="sc">;</span> <span class="k">last</span> <span class="j">SEQFORMAT</span><span class="s">}</span>
|
|
264 251 <span class="i">$Status</span> = <span class="n">0</span><span class="sc">;</span> <span class="i">$SequenceFormat</span> = <span class="q">'NotSupported'</span><span class="sc">;</span>
|
|
265 252 <span class="s">}</span>
|
|
266 253 <span class="k">return</span> <span class="s">(</span><span class="i">$Status</span><span class="cm">,</span> <span class="i">$SequenceFormat</span><span class="s">)</span><span class="sc">;</span>
|
|
267 254 <span class="s">}</span>
|
|
268 255
|
|
269 256 <span class="c"># Is it a ClustalW multiple sequence sequence file...</span>
|
|
270 <a name="IsClustalWSequenceFile-"></a> 257 <span class="k">sub </span><span class="m">IsClustalWSequenceFile</span> <span class="s">{</span>
|
|
271 258 <span class="k">my</span><span class="s">(</span><span class="i">$SequenceFile</span><span class="s">)</span> = <span class="i">@_</span><span class="sc">;</span>
|
|
272 259 <span class="k">my</span><span class="s">(</span><span class="i">$Status</span><span class="cm">,</span> <span class="i">$Line</span><span class="s">)</span><span class="sc">;</span>
|
|
273 260
|
|
274 261 <span class="i">$Status</span> = <span class="n">0</span><span class="sc">;</span>
|
|
275 262
|
|
276 263 <span class="k">open</span> <span class="w">SEQUENCEFILE</span><span class="cm">,</span> <span class="q">"$SequenceFile"</span> <span class="k">or</span> <span class="k">die</span> <span class="q">"Couldn't open $SequenceFile: $!\n"</span><span class="sc">;</span>
|
|
277 264 <span class="i">$Line</span> = <span class="i">GetTextLine</span><span class="s">(</span>\<span class="i">*SEQUENCEFILE</span><span class="s">)</span><span class="sc">;</span>
|
|
278 265 <span class="i">$Status</span> = <span class="s">(</span><span class="i">$Line</span> =~ <span class="q">/(ClustalW|Clustal W|Clustal)/i</span> <span class="s">)</span> ? <span class="n">1</span> <span class="co">:</span> <span class="n">0</span><span class="sc">;</span>
|
|
279 266 <span class="k">close</span> <span class="w">SEQUENCEFILE</span><span class="sc">;</span>
|
|
280 267
|
|
281 268 <span class="k">return</span> <span class="i">$Status</span><span class="sc">;</span>
|
|
282 269 <span class="s">}</span>
|
|
283 270
|
|
284 271 <span class="c"># Is it a valid Pearson fasta sequence or alignment file?</span>
|
|
285 272 <span class="c">#</span>
|
|
286 <a name="IsPearsonFastaSequenceFile-"></a> 273 <span class="k">sub </span><span class="m">IsPearsonFastaSequenceFile</span> <span class="s">{</span>
|
|
287 274 <span class="k">my</span><span class="s">(</span><span class="i">$FastaFile</span><span class="cm">,</span> <span class="i">$Line</span><span class="cm">,</span> <span class="i">$Status</span><span class="s">)</span><span class="sc">;</span>
|
|
288 275
|
|
289 276 <span class="s">(</span><span class="i">$FastaFile</span><span class="s">)</span> = <span class="i">@_</span><span class="sc">;</span>
|
|
290 277 <span class="i">$Status</span> = <span class="n">0</span><span class="sc">;</span>
|
|
291 278
|
|
292 279 <span class="k">open</span> <span class="w">FASTAFILE</span><span class="cm">,</span> <span class="q">"$FastaFile"</span> <span class="k">or</span> <span class="k">die</span> <span class="q">"Couldn't open $FastaFile: $!\n"</span><span class="sc">;</span>
|
|
293 280 <span class="i">$Line</span> = <span class="i">GetTextLine</span><span class="s">(</span>\<span class="i">*FASTAFILE</span><span class="s">)</span><span class="sc">;</span>
|
|
294 281
|
|
295 282 <span class="c"># First line starts with > and the fourth character is not ';'; otherwise, it's</span>
|
|
296 283 <span class="c"># PIR FASTA format.</span>
|
|
297 284 <span class="k">if</span> <span class="s">(</span><span class="i">$Line</span> =~ <span class="q">/^>/</span><span class="s">)</span> <span class="s">{</span>
|
|
298 285 <span class="k">my</span><span class="s">(</span><span class="i">$FourthChar</span><span class="s">)</span><span class="sc">;</span>
|
|
299 286 <span class="i">$FourthChar</span> = <span class="k">substr</span><span class="s">(</span><span class="i">$Line</span><span class="cm">,</span> <span class="n">3</span><span class="cm">,</span> <span class="n">1</span><span class="s">)</span><span class="sc">;</span>
|
|
300 287 <span class="i">$Status</span> = <span class="s">(</span><span class="i">$FourthChar</span> !~ <span class="q">/\;/</span><span class="s">)</span> ? <span class="n">1</span> <span class="co">:</span> <span class="n">0</span><span class="sc">;</span>
|
|
301 288 <span class="s">}</span>
|
|
302 289 <span class="k">close</span> <span class="w">FASTAFILE</span><span class="sc">;</span>
|
|
303 290
|
|
304 291 <span class="k">return</span> <span class="i">$Status</span><span class="sc">;</span>
|
|
305 292 <span class="s">}</span>
|
|
306 293
|
|
307 294 <span class="c"># Is it a valid NBRF/PIR fasta sequence or alignment file?</span>
|
|
308 295 <span class="c">#</span>
|
|
309 <a name="IsPIRFastaSequenceFile-"></a> 296 <span class="k">sub </span><span class="m">IsPIRFastaSequenceFile</span> <span class="s">{</span>
|
|
310 297 <span class="k">my</span><span class="s">(</span><span class="i">$FastaFile</span><span class="cm">,</span> <span class="i">$Line</span><span class="cm">,</span> <span class="i">$Status</span><span class="s">)</span><span class="sc">;</span>
|
|
311 298
|
|
312 299 <span class="s">(</span><span class="i">$FastaFile</span><span class="s">)</span> = <span class="i">@_</span><span class="sc">;</span>
|
|
313 300 <span class="i">$Status</span> = <span class="n">0</span><span class="sc">;</span>
|
|
314 301
|
|
315 302 <span class="k">open</span> <span class="w">FASTAFILE</span><span class="cm">,</span> <span class="q">"$FastaFile"</span> <span class="k">or</span> <span class="k">die</span> <span class="q">"Couldn't open $FastaFile: $!\n"</span><span class="sc">;</span>
|
|
316 303 <span class="i">$Line</span> = <span class="i">GetTextLine</span><span class="s">(</span>\<span class="i">*FASTAFILE</span><span class="s">)</span><span class="sc">;</span>
|
|
317 304
|
|
318 305 <span class="c"># First line starts with > and the fourth character is ';'; otherwise, it's</span>
|
|
319 306 <span class="c"># a Pearson FASTA format.</span>
|
|
320 307 <span class="k">if</span> <span class="s">(</span><span class="i">$Line</span> =~ <span class="q">/^>/</span><span class="s">)</span> <span class="s">{</span>
|
|
321 308 <span class="k">my</span><span class="s">(</span><span class="i">$FourthChar</span><span class="s">)</span><span class="sc">;</span>
|
|
322 309 <span class="i">$FourthChar</span> = <span class="k">substr</span><span class="s">(</span><span class="i">$Line</span><span class="cm">,</span> <span class="n">3</span><span class="cm">,</span> <span class="n">1</span><span class="s">)</span><span class="sc">;</span>
|
|
323 310 <span class="i">$Status</span> = <span class="s">(</span><span class="i">$FourthChar</span> =~ <span class="q">/\;/</span><span class="s">)</span> ? <span class="n">1</span> <span class="co">:</span> <span class="n">0</span><span class="sc">;</span>
|
|
324 311 <span class="s">}</span>
|
|
325 312 <span class="k">close</span> <span class="w">FASTAFILE</span><span class="sc">;</span>
|
|
326 313
|
|
327 314 <span class="k">return</span> <span class="i">$Status</span><span class="sc">;</span>
|
|
328 315 <span class="s">}</span>
|
|
329 316
|
|
330 317 <span class="c"># Is it a valid MSF sequence or alignment file?</span>
|
|
331 318 <span class="c">#</span>
|
|
332 <a name="IsMSFSequenceFile-"></a> 319 <span class="k">sub </span><span class="m">IsMSFSequenceFile</span> <span class="s">{</span>
|
|
333 320 <span class="k">my</span><span class="s">(</span><span class="i">$MSFFile</span><span class="s">)</span> = <span class="i">@_</span><span class="sc">;</span>
|
|
334 321
|
|
335 322 <span class="k">open</span> <span class="w">MSFFILE</span><span class="cm">,</span> <span class="q">"$MSFFile"</span> <span class="k">or</span> <span class="k">die</span> <span class="q">"Couldn't open $MSFFile: $!\n"</span><span class="sc">;</span>
|
|
336 323
|
|
337 324 <span class="k">my</span><span class="s">(</span><span class="i">$Line</span><span class="cm">,</span> <span class="i">$Status</span><span class="s">)</span><span class="sc">;</span>
|
|
338 325
|
|
339 326 <span class="i">$Status</span> = <span class="n">0</span><span class="sc">;</span>
|
|
340 327 <span class="c"># Find a line that contains MSF: keyword and ends with '..'</span>
|
|
341 328 <span class="j">LINE:</span> <span class="k">while</span> <span class="s">(</span><span class="i">$Line</span> = <span class="i">GetTextLine</span><span class="s">(</span>\<span class="i">*MSFFILE</span><span class="s">)</span><span class="s">)</span> <span class="s">{</span>
|
|
342 329 <span class="i">$Line</span> = <span class="i">RemoveLeadingWhiteSpaces</span><span class="s">(</span><span class="i">$Line</span><span class="s">)</span><span class="sc">;</span>
|
|
343 330 <span class="k">if</span> <span class="s">(</span><span class="i">$Line</span> =~ <span class="q">/MSF:/i</span> && <span class="i">$Line</span> =~ <span class="q">/\.\.[ ]*$/</span><span class="s">)</span> <span class="s">{</span>
|
|
344 331 <span class="i">$Status</span> = <span class="n">1</span><span class="sc">;</span>
|
|
345 332 <span class="k">last</span> <span class="j">LINE</span><span class="sc">;</span>
|
|
346 333 <span class="s">}</span>
|
|
347 334 <span class="k">elsif</span> <span class="s">(</span><span class="i">$Line</span> =~ <span class="q">/(!!AA_MULTIPLE_ALIGNMENT|!!NA_MULTIPLE_ALIGNMENT|PILEUP)/i</span><span class="s">)</span> <span class="s">{</span>
|
|
348 335 <span class="c"># Pileup MSF...</span>
|
|
349 336 <span class="i">$Status</span> = <span class="n">1</span><span class="sc">;</span>
|
|
350 337 <span class="k">last</span> <span class="j">LINE</span><span class="sc">;</span>
|
|
351 338 <span class="s">}</span>
|
|
352 339 <span class="s">}</span>
|
|
353 340 <span class="k">close</span> <span class="w">MSFFILE</span><span class="sc">;</span>
|
|
354 341
|
|
355 342 <span class="k">return</span> <span class="i">$Status</span><span class="sc">;</span>
|
|
356 343 <span class="s">}</span>
|
|
357 344
|
|
358 345 <span class="c"># Read sequence or sequence alignment file...</span>
|
|
359 <a name="ReadSequenceFile-"></a> 346 <span class="k">sub </span><span class="m">ReadSequenceFile</span> <span class="s">{</span>
|
|
360 347 <span class="k">my</span><span class="s">(</span><span class="i">$SequenceFile</span><span class="s">)</span> = <span class="i">@_</span><span class="sc">;</span>
|
|
361 348
|
|
362 349 <span class="k">if</span> <span class="s">(</span><span class="i">IsPearsonFastaSequenceFile</span><span class="s">(</span><span class="i">$SequenceFile</span><span class="s">)</span><span class="s">)</span> <span class="s">{</span>
|
|
363 350 <span class="k">return</span> <span class="i">ReadPearsonFastaSequenceFile</span><span class="s">(</span><span class="i">$SequenceFile</span><span class="s">)</span><span class="sc">;</span>
|
|
364 351 <span class="s">}</span>
|
|
365 352 <span class="k">elsif</span> <span class="s">(</span><span class="i">IsPIRFastaSequenceFile</span><span class="s">(</span><span class="i">$SequenceFile</span><span class="s">)</span><span class="s">)</span> <span class="s">{</span>
|
|
366 353 <span class="k">return</span> <span class="i">ReadPIRFastaSequenceFile</span><span class="s">(</span><span class="i">$SequenceFile</span><span class="s">)</span><span class="sc">;</span>
|
|
367 354 <span class="s">}</span>
|
|
368 355 <span class="k">elsif</span> <span class="s">(</span><span class="i">IsMSFSequenceFile</span><span class="s">(</span><span class="i">$SequenceFile</span><span class="s">)</span><span class="s">)</span> <span class="s">{</span>
|
|
369 356 <span class="k">return</span> <span class="i">ReadMSFSequenceFile</span><span class="s">(</span><span class="i">$SequenceFile</span><span class="s">)</span><span class="sc">;</span>
|
|
370 357 <span class="s">}</span>
|
|
371 358 <span class="k">elsif</span> <span class="s">(</span><span class="i">IsClustalWSequenceFile</span><span class="s">(</span><span class="i">$SequenceFile</span><span class="s">)</span><span class="s">)</span> <span class="s">{</span>
|
|
372 359 <span class="k">return</span> <span class="i">ReadClustalWSequenceFile</span><span class="s">(</span><span class="i">$SequenceFile</span><span class="s">)</span><span class="sc">;</span>
|
|
373 360 <span class="s">}</span>
|
|
374 361 <span class="k">else</span> <span class="s">{</span>
|
|
375 362 <span class="k">return</span> <span class="k">undef</span><span class="sc">;</span>
|
|
376 363 <span class="s">}</span>
|
|
377 364 <span class="s">}</span>
|
|
378 365
|
|
379 366 <span class="c"># Read file and setup alignment data...</span>
|
|
380 <a name="ReadClustalWSequenceFile-"></a> 367 <span class="k">sub </span><span class="m">ReadClustalWSequenceFile</span> <span class="s">{</span>
|
|
381 368 <span class="k">my</span><span class="s">(</span><span class="i">$SequenceFile</span><span class="s">)</span> = <span class="i">@_</span><span class="sc">;</span>
|
|
382 369
|
|
383 370 <span class="k">return</span> <span class="i">_ReadFileAndSetupSequencesData</span><span class="s">(</span><span class="i">$SequenceFile</span><span class="cm">,</span> <span class="q">'ClustalW'</span><span class="s">)</span><span class="sc">;</span>
|
|
384 371 <span class="s">}</span>
|
|
385 372
|
|
386 373 <span class="c"># Read file and setup alignment data...</span>
|
|
387 <a name="ReadPearsonFastaSequenceFile-"></a> 374 <span class="k">sub </span><span class="m">ReadPearsonFastaSequenceFile</span> <span class="s">{</span>
|
|
388 375 <span class="k">my</span><span class="s">(</span><span class="i">$SequenceFile</span><span class="s">)</span> = <span class="i">@_</span><span class="sc">;</span>
|
|
389 376
|
|
390 377 <span class="k">return</span> <span class="i">_ReadFileAndSetupSequencesData</span><span class="s">(</span><span class="i">$SequenceFile</span><span class="cm">,</span> <span class="q">'Pearson'</span><span class="s">)</span><span class="sc">;</span>
|
|
391 378 <span class="s">}</span>
|
|
392 379
|
|
393 380 <span class="c"># Read file and setup alignment data...</span>
|
|
394 <a name="ReadPIRFastaSequenceFile-"></a> 381 <span class="k">sub </span><span class="m">ReadPIRFastaSequenceFile</span> <span class="s">{</span>
|
|
395 382 <span class="k">my</span><span class="s">(</span><span class="i">$SequenceFile</span><span class="s">)</span> = <span class="i">@_</span><span class="sc">;</span>
|
|
396 383
|
|
397 384 <span class="k">return</span> <span class="i">_ReadFileAndSetupSequencesData</span><span class="s">(</span><span class="i">$SequenceFile</span><span class="cm">,</span> <span class="q">'PIR'</span><span class="s">)</span><span class="sc">;</span>
|
|
398 385 <span class="s">}</span>
|
|
399 386
|
|
400 387
|
|
401 388 <span class="c"># Read file and setup sequence data...</span>
|
|
402 <a name="ReadMSFSequenceFile-"></a> 389 <span class="k">sub </span><span class="m">ReadMSFSequenceFile</span> <span class="s">{</span>
|
|
403 390 <span class="k">my</span><span class="s">(</span><span class="i">$SequenceFile</span><span class="s">)</span> = <span class="i">@_</span><span class="sc">;</span>
|
|
404 391
|
|
405 392 <span class="k">return</span> <span class="i">_ReadFileAndSetupSequencesData</span><span class="s">(</span><span class="i">$SequenceFile</span><span class="cm">,</span> <span class="q">'MSF'</span><span class="s">)</span><span class="sc">;</span>
|
|
406 393 <span class="s">}</span>
|
|
407 394
|
|
408 395 <span class="c"># Write out a Pearson FASTA file...</span>
|
|
409 <a name="WritePearsonFastaSequenceFile-"></a> 396 <span class="k">sub </span><span class="m">WritePearsonFastaSequenceFile</span> <span class="s">{</span>
|
|
410 397 <span class="k">my</span><span class="s">(</span><span class="i">$SequenceFileName</span><span class="cm">,</span> <span class="i">$SequenceDataRef</span><span class="cm">,</span> <span class="i">$MaxLength</span><span class="cm">,</span> <span class="i">$ID</span><span class="cm">,</span> <span class="i">$Description</span><span class="cm">,</span> <span class="i">$Sequence</span><span class="cm">,</span> <span class="i">$WrappedSequence</span><span class="s">)</span><span class="sc">;</span>
|
|
411 398
|
|
412 399 <span class="i">$MaxLength</span> = <span class="n">80</span><span class="sc">;</span>
|
|
413 400 <span class="k">if</span> <span class="s">(</span><span class="i">@_</span> == <span class="n">3</span><span class="s">)</span> <span class="s">{</span>
|
|
414 401 <span class="s">(</span><span class="i">$SequenceFileName</span><span class="cm">,</span> <span class="i">$SequenceDataRef</span><span class="cm">,</span> <span class="i">$MaxLength</span><span class="s">)</span> = <span class="i">@_</span><span class="sc">;</span>
|
|
415 402 <span class="s">}</span>
|
|
416 403 <span class="k">elsif</span> <span class="s">(</span><span class="i">@_</span> == <span class="n">2</span><span class="s">)</span> <span class="s">{</span>
|
|
417 404 <span class="s">(</span><span class="i">$SequenceFileName</span><span class="cm">,</span> <span class="i">$SequenceDataRef</span><span class="s">)</span> = <span class="i">@_</span><span class="sc">;</span>
|
|
418 405 <span class="s">}</span>
|
|
419 406 <span class="k">open</span> <span class="w">SEQUENCEFILE</span><span class="cm">,</span> <span class="q">">$SequenceFileName"</span> <span class="k">or</span> <span class="k">die</span> <span class="q">"Can't open $SequenceFileName: $!\n"</span><span class="sc">;</span>
|
|
420 407 <span class="k">for</span> <span class="i">$ID</span> <span class="s">(</span><span class="i">@</span>{<span class="i">$SequenceDataRef</span>->{<span class="w">IDs</span>}}<span class="s">)</span> <span class="s">{</span>
|
|
421 408 <span class="i">$Description</span> = <span class="i">$SequenceDataRef</span>->{<span class="w">Description</span>}{<span class="i">$ID</span>}<span class="sc">;</span>
|
|
422 409 <span class="i">$Sequence</span> = <span class="i">$SequenceDataRef</span>->{<span class="w">Sequence</span>}{<span class="i">$ID</span>}<span class="sc">;</span>
|
|
423 410 <span class="i">$WrappedSequence</span> = <span class="i">WrapText</span><span class="s">(</span><span class="i">$Sequence</span><span class="cm">,</span> <span class="i">$MaxLength</span><span class="cm">,</span> <span class="q">"\n"</span><span class="s">)</span><span class="sc">;</span>
|
|
424 411
|
|
425 412 <span class="c"># Description also contains ID...</span>
|
|
426 413 <span class="k">print</span> <span class="i">SEQUENCEFILE</span> <span class="q">">$Description\n"</span><span class="sc">;</span>
|
|
427 414 <span class="k">print</span> <span class="i">SEQUENCEFILE</span> <span class="q">"$WrappedSequence\n"</span><span class="sc">;</span>
|
|
428 415 <span class="s">}</span>
|
|
429 416 <span class="k">close</span> <span class="w">SEQUENCEFILE</span><span class="sc">;</span>
|
|
430 417 <span class="s">}</span>
|
|
431 418
|
|
432 419 <span class="c"># Get ID, Sequence and Length for smallest or longest sequence</span>
|
|
433 <a name="_GetShortestOrLongestSequence-"></a> 420 <span class="k">sub </span><span class="m">_GetShortestOrLongestSequence</span> <span class="s">{</span>
|
|
434 421 <span class="k">my</span><span class="s">(</span><span class="i">$SequencesDataRef</span><span class="cm">,</span> <span class="i">$SequenceType</span><span class="cm">,</span> <span class="i">$IgnoreGaps</span><span class="s">)</span> = <span class="i">@_</span><span class="sc">;</span>
|
|
435 422 <span class="k">my</span><span class="s">(</span><span class="i">$ID</span><span class="cm">,</span> <span class="i">$Seq</span><span class="cm">,</span> <span class="i">$SeqLen</span><span class="cm">,</span> <span class="i">$Description</span><span class="cm">,</span> <span class="i">$FirstID</span><span class="cm">,</span> <span class="i">$FirstSeqLen</span><span class="cm">,</span> <span class="i">$CurrentID</span><span class="cm">,</span> <span class="i">$CurrentSeq</span><span class="cm">,</span> <span class="i">$CurrentSeqLen</span><span class="cm">,</span> <span class="i">$CurrentDescription</span><span class="s">)</span><span class="sc">;</span>
|
|
436 423
|
|
437 424 <span class="s">(</span><span class="i">$ID</span><span class="cm">,</span> <span class="i">$Seq</span><span class="cm">,</span> <span class="i">$SeqLen</span><span class="s">)</span> = <span class="s">(</span><span class="q">''</span><span class="cm">,</span> <span class="q">''</span><span class="cm">,</span> <span class="q">''</span><span class="s">)</span><span class="sc">;</span>
|
|
438 425 <span class="i">$FirstID</span> = <span class="q">''</span><span class="sc">;</span>
|
|
439 426
|
|
440 427 <span class="j">ID:</span> <span class="k">for</span> <span class="i">$CurrentID</span> <span class="s">(</span><span class="i">@</span>{<span class="i">$SequencesDataRef</span>->{<span class="w">IDs</span>}}<span class="s">)</span> <span class="s">{</span>
|
|
441 428 <span class="i">$CurrentSeq</span> = <span class="i">$IgnoreGaps</span> ? <span class="i">RemoveSequenceGaps</span><span class="s">(</span><span class="i">$SequencesDataRef</span>->{<span class="w">Sequence</span>}{<span class="i">$CurrentID</span>}<span class="s">)</span> <span class="co">:</span> <span class="i">$SequencesDataRef</span>->{<span class="w">Sequence</span>}{<span class="i">$CurrentID</span>}<span class="sc">;</span>
|
|
442 429 <span class="i">$CurrentSeqLen</span> = <span class="i">GetSequenceLength</span><span class="s">(</span><span class="i">$CurrentSeq</span><span class="cm">,</span> <span class="i">$IgnoreGaps</span><span class="s">)</span><span class="sc">;</span>
|
|
443 430 <span class="i">$CurrentDescription</span> = <span class="i">$SequencesDataRef</span>->{<span class="w">Description</span>}{<span class="i">$CurrentID</span>}<span class="sc">;</span>
|
|
444 431 <span class="k">if</span> <span class="s">(</span>!<span class="i">$FirstID</span><span class="s">)</span> <span class="s">{</span>
|
|
445 432 <span class="i">$FirstID</span> = <span class="i">$ID</span><span class="sc">;</span> <span class="i">$FirstSeqLen</span> = <span class="i">$CurrentSeqLen</span><span class="sc">;</span>
|
|
446 433 <span class="s">(</span><span class="i">$ID</span><span class="cm">,</span> <span class="i">$Seq</span><span class="cm">,</span> <span class="i">$SeqLen</span><span class="cm">,</span> <span class="i">$Description</span><span class="s">)</span> = <span class="s">(</span><span class="i">$CurrentID</span><span class="cm">,</span> <span class="i">$CurrentSeq</span><span class="cm">,</span> <span class="i">$CurrentSeqLen</span><span class="cm">,</span> <span class="i">$CurrentDescription</span><span class="s">)</span><span class="sc">;</span>
|
|
447 434 <span class="k">next</span> <span class="j">ID</span><span class="sc">;</span>
|
|
448 435 <span class="s">}</span>
|
|
449 436 <span class="k">if</span> <span class="s">(</span><span class="i">$CurrentSeqLen</span> != <span class="i">$SeqLen</span><span class="s">)</span> <span class="s">{</span>
|
|
450 437 <span class="k">if</span> <span class="s">(</span><span class="s">(</span><span class="i">$SequenceType</span> =~ <span class="q">/Shortest/i</span><span class="s">)</span> && <span class="s">(</span><span class="i">$CurrentSeqLen</span> < <span class="i">$SeqLen</span><span class="s">)</span><span class="s">)</span> <span class="s">{</span>
|
|
451 438 <span class="s">(</span><span class="i">$ID</span><span class="cm">,</span> <span class="i">$Seq</span><span class="cm">,</span> <span class="i">$SeqLen</span><span class="cm">,</span> <span class="i">$Description</span><span class="s">)</span> = <span class="s">(</span><span class="i">$CurrentID</span><span class="cm">,</span> <span class="i">$CurrentSeq</span><span class="cm">,</span> <span class="i">$CurrentSeqLen</span><span class="cm">,</span> <span class="i">$CurrentDescription</span><span class="s">)</span><span class="sc">;</span>
|
|
452 439 <span class="s">}</span>
|
|
453 440 <span class="k">elsif</span> <span class="s">(</span><span class="s">(</span><span class="i">$SequenceType</span> =~ <span class="q">/Longest/i</span><span class="s">)</span> && <span class="s">(</span><span class="i">$CurrentSeqLen</span> > <span class="i">$SeqLen</span><span class="s">)</span> <span class="s">)</span> <span class="s">{</span>
|
|
454 441 <span class="s">(</span><span class="i">$ID</span><span class="cm">,</span> <span class="i">$Seq</span><span class="cm">,</span> <span class="i">$SeqLen</span><span class="cm">,</span> <span class="i">$Description</span><span class="s">)</span> = <span class="s">(</span><span class="i">$CurrentID</span><span class="cm">,</span> <span class="i">$CurrentSeq</span><span class="cm">,</span> <span class="i">$CurrentSeqLen</span><span class="cm">,</span> <span class="i">$CurrentDescription</span><span class="s">)</span><span class="sc">;</span>
|
|
455 442 <span class="s">}</span>
|
|
456 443 <span class="s">}</span>
|
|
457 444 <span class="s">}</span>
|
|
458 445 <span class="k">return</span> <span class="s">(</span><span class="i">$ID</span><span class="cm">,</span> <span class="i">$Seq</span><span class="cm">,</span> <span class="i">$SeqLen</span><span class="cm">,</span> <span class="i">$Description</span><span class="s">)</span><span class="sc">;</span>
|
|
459 446 <span class="s">}</span>
|
|
460 447
|
|
461 448 <span class="c"># Remove gaps in the sequence and return new sequence...</span>
|
|
462 <a name="RemoveSequenceGaps-"></a> 449 <span class="k">sub </span><span class="m">RemoveSequenceGaps</span> <span class="s">{</span>
|
|
463 450 <span class="k">my</span><span class="s">(</span><span class="i">$Seq</span><span class="s">)</span> = <span class="i">@_</span><span class="sc">;</span>
|
|
464 451 <span class="k">my</span><span class="s">(</span><span class="i">$SeqWithoutGaps</span><span class="cm">,</span> <span class="i">$SeqLen</span><span class="cm">,</span> <span class="i">$Index</span><span class="cm">,</span> <span class="i">$Residue</span><span class="s">)</span><span class="sc">;</span>
|
|
465 452
|
|
466 453 <span class="i">$SeqWithoutGaps</span> = <span class="q">''</span><span class="sc">;</span>
|
|
467 454 <span class="i">$SeqLen</span> = <span class="k">length</span><span class="s">(</span><span class="i">$Seq</span><span class="s">)</span><span class="sc">;</span>
|
|
468 455 <span class="k">for</span> <span class="i">$Index</span> <span class="s">(</span><span class="n">0</span> .. <span class="s">(</span><span class="i">$SeqLen</span> - <span class="n">1</span><span class="s">)</span><span class="s">)</span> <span class="s">{</span>
|
|
469 456 <span class="i">$Residue</span> = <span class="k">substr</span><span class="s">(</span><span class="i">$Seq</span><span class="cm">,</span> <span class="i">$Index</span><span class="cm">,</span> <span class="n">1</span><span class="s">)</span><span class="sc">;</span>
|
|
470 457 <span class="k">if</span> <span class="s">(</span><span class="i">$Residue</span> =~ <span class="q">/[A-Z]/i</span><span class="s">)</span> <span class="s">{</span>
|
|
471 458 <span class="i">$SeqWithoutGaps</span> .= <span class="i">$Residue</span><span class="sc">;</span>
|
|
472 459 <span class="s">}</span>
|
|
473 460 <span class="s">}</span>
|
|
474 461
|
|
475 462 <span class="k">return</span> <span class="i">$SeqWithoutGaps</span><span class="sc">;</span>
|
|
476 463 <span class="s">}</span>
|
|
477 464
|
|
478 465 <span class="c"># Using input alignment data map ref containing following keys, generate</span>
|
|
479 466 <span class="c"># a new hash with same set of keys after residue columns containg only</span>
|
|
480 467 <span class="c"># gaps have been removed:</span>
|
|
481 468 <span class="c">#</span>
|
|
482 469 <span class="c"># {IDs} : Array of IDs in order as they appear in file</span>
|
|
483 470 <span class="c"># {Count}: ID count...</span>
|
|
484 471 <span class="c"># {Description}{$ID} : Description data...</span>
|
|
485 472 <span class="c"># {Sequence}{$ID} : Sequence data...</span>
|
|
486 473 <span class="c">#</span>
|
|
487 <a name="RemoveSequenceAlignmentGapColumns-"></a> 474 <span class="k">sub </span><span class="m">RemoveSequenceAlignmentGapColumns</span> <span class="s">{</span>
|
|
488 475 <span class="k">my</span><span class="s">(</span><span class="i">$ID</span><span class="cm">,</span> <span class="i">$AlignmentDataMapRef</span><span class="cm">,</span> <span class="i">%NewAlignmentDataMap</span><span class="s">)</span><span class="sc">;</span>
|
|
489 476
|
|
490 477 <span class="s">(</span><span class="i">$AlignmentDataMapRef</span><span class="s">)</span> = <span class="i">@_</span><span class="sc">;</span>
|
|
491 478
|
|
492 479 <span class="i">%NewAlignmentDataMap</span> = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
493 480 <span class="i">@</span>{<span class="i">$NewAlignmentDataMap</span>{<span class="w">IDs</span>}} =<span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
494 481 <span class="i">%</span>{<span class="i">$NewAlignmentDataMap</span>{<span class="w">Description</span>}} =<span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
495 482 <span class="i">%</span>{<span class="i">$NewAlignmentDataMap</span>{<span class="w">Sequence</span>}} =<span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
496 483 <span class="i">$NewAlignmentDataMap</span>{<span class="w">Count</span>} = <span class="n">0</span><span class="sc">;</span>
|
|
497 484
|
|
498 485 <span class="c"># Transfer ID and count information...</span>
|
|
499 486 <span class="k">for</span> <span class="i">$ID</span> <span class="s">(</span><span class="i">@</span>{<span class="i">$AlignmentDataMapRef</span>->{<span class="w">IDs</span>}}<span class="s">)</span> <span class="s">{</span>
|
|
500 487 <span class="k">push</span> <span class="i">@</span>{<span class="i">$NewAlignmentDataMap</span>{<span class="w">IDs</span>}}<span class="cm">,</span> <span class="i">$ID</span><span class="sc">;</span>
|
|
501 488 <span class="i">$NewAlignmentDataMap</span>{<span class="w">Description</span>}{<span class="i">$ID</span>} = <span class="i">$AlignmentDataMapRef</span>->{<span class="w">Description</span>}{<span class="i">$ID</span>}<span class="sc">;</span>
|
|
502 489 <span class="i">$NewAlignmentDataMap</span>{<span class="w">Sequence</span>}{<span class="i">$ID</span>} = <span class="q">''</span><span class="sc">;</span>
|
|
503 490 <span class="i">$NewAlignmentDataMap</span>{<span class="w">Count</span>} += <span class="n">1</span><span class="sc">;</span>
|
|
504 491 <span class="s">}</span>
|
|
505 492
|
|
506 493 <span class="c"># Go over residue columns and transfer the data...</span>
|
|
507 494 <span class="k">my</span><span class="s">(</span><span class="i">$FirstID</span><span class="cm">,</span> <span class="i">$FirstSeq</span><span class="cm">,</span> <span class="i">$FirstSeqLen</span><span class="cm">,</span> <span class="i">$Index</span><span class="cm">,</span> <span class="i">$Res</span><span class="cm">,</span> <span class="i">$GapColumn</span><span class="s">)</span><span class="sc">;</span>
|
|
508 495
|
|
509 496 <span class="i">$FirstID</span> = <span class="i">$AlignmentDataMapRef</span>->{<span class="w">IDs</span>}[<span class="n">0</span>]<span class="sc">;</span>
|
|
510 497 <span class="i">$FirstSeq</span> = <span class="i">$AlignmentDataMapRef</span>->{<span class="w">Sequence</span>}{<span class="i">$FirstID</span>}<span class="sc">;</span>
|
|
511 498 <span class="i">$FirstSeqLen</span> = <span class="k">length</span><span class="s">(</span><span class="i">$FirstSeq</span><span class="s">)</span><span class="sc">;</span>
|
|
512 499
|
|
513 500 <span class="j">RES:</span> <span class="k">for</span> <span class="i">$Index</span> <span class="s">(</span><span class="n">0</span> .. <span class="s">(</span><span class="i">$FirstSeqLen</span> - <span class="n">1</span><span class="s">)</span><span class="s">)</span> <span class="s">{</span>
|
|
514 501 <span class="c"># Is this a gap column?</span>
|
|
515 502 <span class="i">$GapColumn</span> = <span class="n">1</span><span class="sc">;</span>
|
|
516 503 <span class="j">ID:</span> <span class="k">for</span> <span class="i">$ID</span> <span class="s">(</span><span class="i">@</span>{<span class="i">$AlignmentDataMapRef</span>->{<span class="w">IDs</span>}}<span class="s">)</span> <span class="s">{</span>
|
|
517 504 <span class="i">$Res</span> = <span class="k">substr</span><span class="s">(</span><span class="i">$AlignmentDataMapRef</span>->{<span class="w">Sequence</span>}{<span class="i">$ID</span>}<span class="cm">,</span> <span class="i">$Index</span><span class="cm">,</span> <span class="n">1</span><span class="s">)</span><span class="sc">;</span>
|
|
518 505 <span class="k">if</span> <span class="s">(</span><span class="i">$Res</span> =~ <span class="q">/[A-Z]/i</span><span class="s">)</span> <span class="s">{</span>
|
|
519 506 <span class="i">$GapColumn</span> = <span class="n">0</span><span class="sc">;</span>
|
|
520 507 <span class="k">last</span> <span class="j">ID</span><span class="sc">;</span>
|
|
521 508 <span class="s">}</span>
|
|
522 509 <span class="s">}</span>
|
|
523 510 <span class="k">if</span> <span class="s">(</span><span class="i">$GapColumn</span><span class="s">)</span> <span class="s">{</span>
|
|
524 511 <span class="k">next</span> <span class="j">RES</span><span class="sc">;</span>
|
|
525 512 <span class="s">}</span>
|
|
526 513 <span class="c"># Transfer this residue...</span>
|
|
527 514 <span class="k">for</span> <span class="i">$ID</span> <span class="s">(</span><span class="i">@</span>{<span class="i">$AlignmentDataMapRef</span>->{<span class="w">IDs</span>}}<span class="s">)</span> <span class="s">{</span>
|
|
528 515 <span class="i">$Res</span> = <span class="k">substr</span><span class="s">(</span><span class="i">$AlignmentDataMapRef</span>->{<span class="w">Sequence</span>}{<span class="i">$ID</span>}<span class="cm">,</span> <span class="i">$Index</span><span class="cm">,</span> <span class="n">1</span><span class="s">)</span><span class="sc">;</span>
|
|
529 516 <span class="i">$NewAlignmentDataMap</span>{<span class="w">Sequence</span>}{<span class="i">$ID</span>} .= <span class="i">$Res</span><span class="sc">;</span>
|
|
530 517 <span class="s">}</span>
|
|
531 518 <span class="s">}</span>
|
|
532 519
|
|
533 520 <span class="k">return</span> <span class="s">(</span>\<span class="i">%NewAlignmentDataMap</span><span class="s">)</span><span class="sc">;</span>
|
|
534 521 <span class="s">}</span>
|
|
535 522
|
|
536 523 <span class="c">#</span>
|
|
537 524 <span class="c"># Read sequences file and return a reference to hash with the following keys:</span>
|
|
538 525 <span class="c">#</span>
|
|
539 526 <span class="c"># {IDs} - Array of sequence IDs</span>
|
|
540 527 <span class="c"># {Count} - Number of sequences</span>
|
|
541 528 <span class="c"># {Description}{$ID} - Sequence description</span>
|
|
542 529 <span class="c"># {Sequence}{$ID} - Sequence for a specific ID</span>
|
|
543 530 <span class="c"># {InputFileType} - Sequence file format</span>
|
|
544 531 <span class="c"># {ConservedAnnotation} - Conserved residue annonation</span>
|
|
545 532 <span class="c">#</span>
|
|
546 533 <span class="c"># Note:</span>
|
|
547 534 <span class="c"># . Conserved residue annotation either exist in the input sequence alignment file or set</span>
|
|
548 535 <span class="c"># for a file containing same number of residues for all the sequence using the following</span>
|
|
549 536 <span class="c"># notation: * - Residue conserved; ' ' - Residue not conserved.</span>
|
|
550 537 <span class="c">#</span>
|
|
551 <a name="_ReadFileAndSetupSequencesData-"></a> 538 <span class="k">sub </span><span class="m">_ReadFileAndSetupSequencesData</span> <span class="s">{</span>
|
|
552 539 <span class="k">my</span><span class="s">(</span><span class="i">$SequenceFile</span><span class="cm">,</span> <span class="i">$SequenceType</span><span class="s">)</span> = <span class="i">@_</span><span class="sc">;</span>
|
|
553 540 <span class="k">my</span><span class="s">(</span><span class="i">$SequenceDataMapRef</span><span class="s">)</span><span class="sc">;</span>
|
|
554 541
|
|
555 542 <span class="i">$SequenceDataMapRef</span> = <span class="k">undef</span><span class="sc">;</span>
|
|
556 543
|
|
557 544 <span class="c"># Read sequence file...</span>
|
|
558 545 <span class="i">$SequenceDataMapRef</span> = <span class="q">''</span><span class="sc">;</span>
|
|
559 546 <span class="k">if</span> <span class="s">(</span><span class="i">$SequenceType</span> =~ <span class="q">/^ClustalW$/i</span><span class="s">)</span> <span class="s">{</span>
|
|
560 547 <span class="i">$SequenceDataMapRef</span> = <span class="i">_ReadClustalWFile</span><span class="s">(</span><span class="i">$SequenceFile</span><span class="s">)</span><span class="sc">;</span>
|
|
561 548 <span class="s">}</span>
|
|
562 549 <span class="k">elsif</span> <span class="s">(</span><span class="i">$SequenceType</span> =~ <span class="q">/^Pearson$/i</span><span class="s">)</span> <span class="s">{</span>
|
|
563 550 <span class="i">$SequenceDataMapRef</span> = <span class="i">_ReadPearsonFastaFile</span><span class="s">(</span><span class="i">$SequenceFile</span><span class="s">)</span><span class="sc">;</span>
|
|
564 551 <span class="s">}</span>
|
|
565 552 <span class="k">elsif</span> <span class="s">(</span><span class="i">$SequenceType</span> =~ <span class="q">/^PIR$/i</span><span class="s">)</span> <span class="s">{</span>
|
|
566 553 <span class="i">$SequenceDataMapRef</span> = <span class="i">_ReadPIRFastaFile</span><span class="s">(</span><span class="i">$SequenceFile</span><span class="s">)</span><span class="sc">;</span>
|
|
567 554 <span class="s">}</span>
|
|
568 555 <span class="k">elsif</span> <span class="s">(</span><span class="i">$SequenceType</span> =~ <span class="q">/^MSF$/i</span><span class="s">)</span> <span class="s">{</span>
|
|
569 556 <span class="i">$SequenceDataMapRef</span> = <span class="i">_ReadMSFFile</span><span class="s">(</span><span class="i">$SequenceFile</span><span class="s">)</span><span class="sc">;</span>
|
|
570 557 <span class="s">}</span>
|
|
571 558 <span class="k">else</span> <span class="s">{</span>
|
|
572 559 <span class="k">return</span> <span class="i">$SequenceDataMapRef</span><span class="sc">;</span>
|
|
573 560 <span class="s">}</span>
|
|
574 561
|
|
575 562 <span class="k">if</span> <span class="s">(</span><span class="k">exists</span> <span class="i">$SequenceDataMapRef</span>->{<span class="w">ConservedAnnotation</span>}<span class="s">)</span> <span class="s">{</span>
|
|
576 563 <span class="k">return</span> <span class="s">(</span><span class="i">$SequenceDataMapRef</span><span class="s">)</span><span class="sc">;</span>
|
|
577 564 <span class="s">}</span>
|
|
578 565 <span class="k">if</span> <span class="s">(</span>!<span class="s">(</span><span class="s">(</span><span class="i">$SequenceDataMapRef</span>->{<span class="w">Count</span>} > <span class="n">1</span><span class="s">)</span> && <span class="s">(</span><span class="i">AreSequenceLengthsIdentical</span><span class="s">(</span><span class="i">$SequenceDataMapRef</span><span class="s">)</span><span class="s">)</span><span class="s">)</span><span class="s">)</span> <span class="s">{</span>
|
|
579 566 <span class="k">return</span> <span class="s">(</span><span class="i">$SequenceDataMapRef</span><span class="s">)</span><span class="sc">;</span>
|
|
580 567 <span class="s">}</span>
|
|
581 568
|
|
582 569 <span class="c"># Use the first sequence to setup an empty ConservedAnnotation key...</span>
|
|
583 570 <span class="c"># And mark fully conserved residues...</span>
|
|
584 571 <span class="c">#</span>
|
|
585 572 <span class="k">my</span><span class="s">(</span><span class="i">$ID</span><span class="cm">,</span> <span class="i">$Sequence</span><span class="cm">,</span> <span class="i">$FirstSequence</span><span class="cm">,</span> <span class="i">$FirstSeqLen</span><span class="cm">,</span> <span class="i">$Res</span><span class="cm">,</span> <span class="i">$FirstRes</span><span class="cm">,</span> <span class="i">$ResConserved</span><span class="cm">,</span> <span class="i">$Index</span><span class="s">)</span><span class="sc">;</span>
|
|
586 573 <span class="i">$ID</span> = <span class="i">$SequenceDataMapRef</span>->{<span class="w">IDs</span>}[<span class="n">0</span>]<span class="sc">;</span>
|
|
587 574 <span class="i">$FirstSequence</span> = <span class="i">$SequenceDataMapRef</span>->{<span class="w">Sequence</span>}{<span class="i">$ID</span>}<span class="sc">;</span>
|
|
588 575 <span class="i">$FirstSeqLen</span> = <span class="k">length</span><span class="s">(</span><span class="i">$FirstSequence</span><span class="s">)</span><span class="sc">;</span>
|
|
589 576 <span class="i">$SequenceDataMapRef</span>->{<span class="w">ConservedAnnotation</span>} = <span class="q">''</span><span class="sc">;</span>
|
|
590 577 <span class="k">for</span> <span class="i">$Index</span> <span class="s">(</span><span class="n">0</span> .. <span class="s">(</span><span class="i">$FirstSeqLen</span> - <span class="n">1</span><span class="s">)</span><span class="s">)</span> <span class="s">{</span>
|
|
591 578 <span class="i">$FirstRes</span> = <span class="q">''</span><span class="sc">;</span>
|
|
592 579 <span class="i">$ResConserved</span> = <span class="n">1</span><span class="sc">;</span>
|
|
593 580 <span class="j">ID:</span> <span class="k">for</span> <span class="i">$ID</span> <span class="s">(</span><span class="i">@</span>{<span class="i">$SequenceDataMapRef</span>->{<span class="w">IDs</span>}}<span class="s">)</span> <span class="s">{</span>
|
|
594 581 <span class="i">$Sequence</span> = <span class="i">$SequenceDataMapRef</span>->{<span class="w">Sequence</span>}{<span class="i">$ID</span>}<span class="sc">;</span>
|
|
595 582 <span class="i">$Res</span> = <span class="k">substr</span><span class="s">(</span><span class="i">$Sequence</span><span class="cm">,</span> <span class="i">$Index</span><span class="cm">,</span> <span class="n">1</span><span class="s">)</span><span class="sc">;</span>
|
|
596 583 <span class="k">if</span> <span class="s">(</span>!<span class="i">$FirstRes</span><span class="s">)</span> <span class="s">{</span>
|
|
597 584 <span class="i">$FirstRes</span> = <span class="i">$Res</span><span class="sc">;</span>
|
|
598 585 <span class="k">next</span> <span class="j">ID</span><span class="sc">;</span>
|
|
599 586 <span class="s">}</span>
|
|
600 587 <span class="k">if</span> <span class="s">(</span><span class="s">(</span><span class="i">$Res</span> !~ <span class="q">/[A-Z]/i</span><span class="s">)</span> || <span class="s">(</span><span class="i">$Res</span> <span class="k">ne</span> <span class="i">$FirstRes</span><span class="s">)</span><span class="s">)</span> <span class="s">{</span>
|
|
601 588 <span class="i">$ResConserved</span> = <span class="n">0</span><span class="sc">;</span>
|
|
602 589 <span class="k">last</span> <span class="j">ID</span><span class="sc">;</span>
|
|
603 590 <span class="s">}</span>
|
|
604 591 <span class="s">}</span>
|
|
605 592 <span class="k">if</span> <span class="s">(</span><span class="i">$ResConserved</span><span class="s">)</span> <span class="s">{</span>
|
|
606 593 <span class="i">$SequenceDataMapRef</span>->{<span class="w">ConservedAnnotation</span>} .= <span class="q">'*'</span><span class="sc">;</span>
|
|
607 594 <span class="s">}</span>
|
|
608 595 <span class="k">else</span> <span class="s">{</span>
|
|
609 596 <span class="i">$SequenceDataMapRef</span>->{<span class="w">ConservedAnnotation</span>} .= <span class="q">' '</span><span class="sc">;</span>
|
|
610 597 <span class="s">}</span>
|
|
611 598 <span class="s">}</span>
|
|
612 599
|
|
613 600 <span class="k">return</span> <span class="s">(</span><span class="i">$SequenceDataMapRef</span><span class="s">)</span><span class="sc">;</span>
|
|
614 601 <span class="s">}</span>
|
|
615 602
|
|
616 603 <span class="c"># Read sequence data in ClustalW multiple sequence alignment file and</span>
|
|
617 604 <span class="c"># return a reference to hash with these keys and values:</span>
|
|
618 605 <span class="c">#</span>
|
|
619 606 <span class="c"># {IDs} - Array of sequence IDs</span>
|
|
620 607 <span class="c"># {Count} - Number of sequences</span>
|
|
621 608 <span class="c"># {Description}{$ID} - Sequence description</span>
|
|
622 609 <span class="c"># {Sequence}{$ID} - Sequence for a specific ID</span>
|
|
623 610 <span class="c"># {InputFileType} - Sequence file format</span>
|
|
624 611 <span class="c"># {ConservedAnnotation} - Conserved residue annonations: space, *, : , .</span>
|
|
625 612 <span class="c">#</span>
|
|
626 613 <span class="c">#</span>
|
|
627 614 <span class="c">#</span>
|
|
628 615 <span class="c"># And based on ClustalW/X manual, here is what the ConservedAnnonations mean:</span>
|
|
629 616 <span class="c">#</span>
|
|
630 617 <span class="c"># '*' indicates positions which have a single, fully conserved residue</span>
|
|
631 618 <span class="c">#</span>
|
|
632 619 <span class="c"># ':' indicates that one of the following 'strong' groups is fully conserved: STA</span>
|
|
633 620 <span class="c"># NEQK NHQK NDEQ QHRK MILV MILF HY FYW</span>
|
|
634 621
|
|
635 622 <span class="c"># '.' indicates that one of the following 'weaker' groups is fully conserved:</span>
|
|
636 623 <span class="c"># CSA ATV SAG STNK STPA SGND SNDEQK NDEQHK NEQHRK FVLIM HFY</span>
|
|
637 624 <span class="c">#</span>
|
|
638 625 <span class="c"># These are all the positively scoring groups that occur in the Gonnet Pam250</span>
|
|
639 626 <span class="c"># matrix. The strong and weak groups are defined as strong score >0.5 and weak</span>
|
|
640 627 <span class="c"># score =<0.5 respectively.</span>
|
|
641 628 <span class="c">#</span>
|
|
642 <a name="_ReadClustalWFile-"></a> 629 <span class="k">sub </span><span class="m">_ReadClustalWFile</span> <span class="s">{</span>
|
|
643 630 <span class="k">my</span><span class="s">(</span><span class="i">$SequenceFile</span><span class="s">)</span> = <span class="i">@_</span><span class="sc">;</span>
|
|
644 631 <span class="k">my</span><span class="s">(</span><span class="i">%SequencesDataMap</span><span class="s">)</span><span class="sc">;</span>
|
|
645 632
|
|
646 633 <span class="c"># Initialize data...</span>
|
|
647 634 <span class="i">%SequencesDataMap</span> = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
648 635 <span class="i">@</span>{<span class="i">$SequencesDataMap</span>{<span class="w">IDs</span>}} = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
649 636 <span class="i">%</span>{<span class="i">$SequencesDataMap</span>{<span class="w">Description</span>}} = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
650 637 <span class="i">%</span>{<span class="i">$SequencesDataMap</span>{<span class="w">Sequence</span>}} = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
651 638 <span class="i">$SequencesDataMap</span>{<span class="w">Count</span>} = <span class="n">0</span><span class="sc">;</span>
|
|
652 639 <span class="i">$SequencesDataMap</span>{<span class="w">ConservedAnnotation</span>} = <span class="q">''</span><span class="sc">;</span>
|
|
653 640 <span class="i">$SequencesDataMap</span>{<span class="w">InputFileType</span>} = <span class="q">'ClustalW'</span><span class="sc">;</span>
|
|
654 641
|
|
655 642 <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">"Couldn't open $SequenceFile: $!\n"</span><span class="sc">;</span>
|
|
656 643
|
|
657 644 <span class="k">my</span><span class="s">(</span><span class="i">$Line</span><span class="cm">,</span> <span class="i">$LineLength</span><span class="cm">,</span> <span class="i">$AnnotationStart</span><span class="cm">,</span> <span class="i">$AnnotationLength</span><span class="cm">,</span> <span class="i">$Annotation</span><span class="cm">,</span> <span class="i">$Sequence</span><span class="cm">,</span> <span class="i">$SequenceLength</span><span class="cm">,</span> <span class="i">$ID</span><span class="cm">,</span> <span class="i">$IDIndex</span><span class="s">)</span><span class="sc">;</span>
|
|
658 645
|
|
659 646 <span class="c"># Ignore the header line...</span>
|
|
660 647 <span class="i">$Line</span> = <span class="q"><SEQUENCEFILE></span><span class="sc">;</span>
|
|
661 648
|
|
662 649 <span class="j">LINE:</span> <span class="k">while</span> <span class="s">(</span><span class="i">$Line</span> = <span class="i">GetTextLine</span><span class="s">(</span>\<span class="i">*SEQUENCEFILE</span><span class="s">)</span><span class="s">)</span> <span class="s">{</span>
|
|
663 650 <span class="k">if</span> <span class="s">(</span><span class="s">(</span><span class="i">$Line</span> =~ <span class="q">/^[ \*\:\.]/</span><span class="s">)</span> && <span class="s">(</span><span class="i">$Line</span> !~ <span class="q">/[A-Z]/i</span><span class="s">)</span><span class="s">)</span> <span class="s">{</span>
|
|
664 651 <span class="c"># Annotation for sequences: fully conserverd, weaker or stronger group conserverd.</span>
|
|
665 652 <span class="c"># Extract it and save...</span>
|
|
666 653 <span class="i">$LineLength</span> = <span class="k">length</span><span class="s">(</span><span class="i">$Line</span><span class="s">)</span><span class="sc">;</span>
|
|
667 654 <span class="i">$AnnotationStart</span> = <span class="i">$LineLength</span> - <span class="i">$SequenceLength</span><span class="sc">;</span>
|
|
668 655 <span class="i">$AnnotationLength</span> = <span class="i">$SequenceLength</span><span class="sc">;</span>
|
|
669 656 <span class="i">$Annotation</span> = <span class="k">substr</span><span class="s">(</span><span class="i">$Line</span><span class="cm">,</span> <span class="i">$AnnotationStart</span><span class="cm">,</span> <span class="i">$AnnotationLength</span><span class="s">)</span><span class="sc">;</span>
|
|
670 657 <span class="i">$SequencesDataMap</span>{<span class="w">ConservedAnnotation</span>} .= <span class="i">$Annotation</span><span class="sc">;</span>
|
|
671 658 <span class="s">}</span>
|
|
672 659 <span class="k">else</span> <span class="s">{</span>
|
|
673 660 <span class="c"># Extract ID and sequences...</span>
|
|
674 661 <span class="s">(</span><span class="i">$ID</span><span class="cm">,</span> <span class="i">$Sequence</span><span class="s">)</span>= <span class="i">$Line</span> =~ <span class="q">/^[ ]*(.*?)[ ]+(.*?)[ 01-9]*$/</span><span class="sc">;</span>
|
|
675 662 <span class="i">$Sequence</span> =~ <span class="q">s/ //g</span><span class="sc">;</span>
|
|
676 663 <span class="k">if</span> <span class="s">(</span>!<span class="s">(</span><span class="i">$ID</span> && <span class="i">$Sequence</span><span class="s">)</span><span class="s">)</span> <span class="s">{</span>
|
|
677 664 <span class="k">next</span> <span class="j">LINE</span><span class="sc">;</span>
|
|
678 665 <span class="s">}</span>
|
|
679 666
|
|
680 667 <span class="k">if</span> <span class="s">(</span><span class="k">exists</span> <span class="i">$SequencesDataMap</span>{<span class="w">Sequence</span>}{<span class="i">$ID</span>}<span class="s">)</span> <span class="s">{</span>
|
|
681 668 <span class="c"># Append to existing alignment value...</span>
|
|
682 669 <span class="i">$SequenceLength</span> = <span class="k">length</span><span class="s">(</span><span class="i">$Sequence</span><span class="s">)</span><span class="sc">;</span>
|
|
683 670 <span class="i">$SequencesDataMap</span>{<span class="w">Sequence</span>}{<span class="i">$ID</span>} .= <span class="i">$Sequence</span><span class="sc">;</span>
|
|
684 671 <span class="s">}</span>
|
|
685 672 <span class="k">else</span> <span class="s">{</span>
|
|
686 673 <span class="c"># New alignment data...</span>
|
|
687 674 <span class="i">$SequencesDataMap</span>{<span class="w">Count</span>} += <span class="n">1</span><span class="sc">;</span>
|
|
688 675 <span class="k">push</span> <span class="i">@</span>{<span class="i">$SequencesDataMap</span>{<span class="w">IDs</span>}}<span class="cm">,</span> <span class="i">$ID</span><span class="sc">;</span>
|
|
689 676 <span class="i">$SequencesDataMap</span>{<span class="w">Description</span>}{<span class="i">$ID</span>} = <span class="i">$ID</span><span class="sc">;</span>
|
|
690 677 <span class="i">$SequencesDataMap</span>{<span class="w">Sequence</span>}{<span class="i">$ID</span>} = <span class="i">$Sequence</span><span class="sc">;</span>
|
|
691 678 <span class="i">$SequenceLength</span> = <span class="k">length</span><span class="s">(</span><span class="i">$Sequence</span><span class="s">)</span><span class="sc">;</span>
|
|
692 679 <span class="s">}</span>
|
|
693 680 <span class="s">}</span>
|
|
694 681 <span class="s">}</span>
|
|
695 682 <span class="k">close</span> <span class="w">SEQUENCEFILE</span><span class="sc">;</span>
|
|
696 683 <span class="k">return</span> <span class="s">(</span>\<span class="i">%SequencesDataMap</span><span class="s">)</span><span class="sc">;</span>
|
|
697 684 <span class="s">}</span>
|
|
698 685
|
|
699 686 <span class="c"># Read Pearson fasta file and return a reference to hash with these keys:</span>
|
|
700 687 <span class="c">#</span>
|
|
701 688 <span class="c"># {IDs} - Array of sequence IDs</span>
|
|
702 689 <span class="c"># {Count} - Number of sequences</span>
|
|
703 690 <span class="c"># {Description}{$ID} - Sequence description</span>
|
|
704 691 <span class="c"># {Sequence}{$ID} - Sequence for a specific ID</span>
|
|
705 692 <span class="c"># {InputFileType} - Sequence file format</span>
|
|
706 693 <span class="c"># {ConservedAnnotation} - Conserved residue annonation</span>
|
|
707 694 <span class="c">#</span>
|
|
708 <a name="_ReadPearsonFastaFile-"></a> 695 <span class="k">sub </span><span class="m">_ReadPearsonFastaFile</span> <span class="s">{</span>
|
|
709 696 <span class="k">my</span><span class="s">(</span><span class="i">$FastaFileName</span><span class="cm">,</span> <span class="i">$ID</span><span class="cm">,</span> <span class="i">$Description</span><span class="cm">,</span> <span class="i">$Line</span><span class="cm">,</span> <span class="i">$IgnoreID</span><span class="cm">,</span> <span class="i">@LineWords</span><span class="cm">,</span> <span class="i">%FastaDataMap</span><span class="s">)</span><span class="sc">;</span>
|
|
710 697
|
|
711 698 <span class="s">(</span><span class="i">$FastaFileName</span><span class="s">)</span> = <span class="i">@_</span><span class="sc">;</span>
|
|
712 699
|
|
713 700 <span class="i">%FastaDataMap</span> = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
714 701 <span class="i">@</span>{<span class="i">$FastaDataMap</span>{<span class="w">IDs</span>}} =<span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
715 702 <span class="i">%</span>{<span class="i">$FastaDataMap</span>{<span class="w">Description</span>}} =<span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
716 703 <span class="i">%</span>{<span class="i">$FastaDataMap</span>{<span class="w">Sequence</span>}} =<span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
717 704 <span class="i">$FastaDataMap</span>{<span class="w">Count</span>} = <span class="n">0</span><span class="sc">;</span>
|
|
718 705 <span class="i">$FastaDataMap</span>{<span class="w">InputFileType</span>} = <span class="q">'Pearson'</span><span class="sc">;</span>
|
|
719 706
|
|
720 707 <span class="k">open</span> <span class="w">FASTAFILE</span><span class="cm">,</span> <span class="q">"$FastaFileName"</span> <span class="k">or</span> <span class="k">die</span> <span class="q">"Couldn't open $FastaFileName: $!\n"</span><span class="sc">;</span>
|
|
721 708 <span class="i">$ID</span> = <span class="q">''</span><span class="sc">;</span>
|
|
722 709 <span class="i">$IgnoreID</span> = <span class="n">0</span><span class="sc">;</span>
|
|
723 710 <span class="j">LINE:</span> <span class="k">while</span> <span class="s">(</span><span class="i">$Line</span> = <span class="i">GetTextLine</span><span class="s">(</span>\<span class="i">*FASTAFILE</span><span class="s">)</span><span class="s">)</span> <span class="s">{</span>
|
|
724 711 <span class="k">if</span> <span class="s">(</span><span class="i">$Line</span> =~ <span class="q">/^\>/</span><span class="s">)</span> <span class="s">{</span>
|
|
725 712 <span class="c"># Start of a new ID...</span>
|
|
726 713 <span class="i">$Line</span> =~ <span class="q">s/^\>//</span><span class="sc">;</span>
|
|
727 714 <span class="i">$Line</span> = <span class="i">RemoveLeadingWhiteSpaces</span><span class="s">(</span><span class="i">$Line</span><span class="s">)</span><span class="sc">;</span>
|
|
728 715 <span class="i">@LineWords</span> = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
729 716 <span class="i">@LineWords</span> = <span class="k">split</span> <span class="q">/ /</span><span class="cm">,</span> <span class="i">$Line</span><span class="sc">;</span>
|
|
730 717
|
|
731 718 <span class="i">$ID</span> = <span class="i">$LineWords</span>[<span class="n">0</span>]<span class="sc">;</span>
|
|
732 719 <span class="i">$ID</span> =~ <span class="q">s/ //g</span><span class="sc">;</span>
|
|
733 720 <span class="i">$Description</span> = <span class="i">$Line</span><span class="sc">;</span>
|
|
734 721
|
|
735 722 <span class="i">$IgnoreID</span> = <span class="n">0</span><span class="sc">;</span>
|
|
736 723 <span class="k">if</span> <span class="s">(</span><span class="k">exists</span> <span class="i">$FastaDataMap</span>{<span class="w">Sequence</span>}{<span class="i">$ID</span>}<span class="s">)</span> <span class="s">{</span>
|
|
737 724 <span class="i">$IgnoreID</span> = <span class="n">1</span><span class="sc">;</span>
|
|
738 725 <span class="k">warn</span> <span class="q">"Warning: ID, $ID, in Fasta file already exists. Ignoring ID and sequence data...\n"</span><span class="sc">;</span>
|
|
739 726 <span class="k">next</span> <span class="j">LINE</span><span class="sc">;</span>
|
|
740 727 <span class="s">}</span>
|
|
741 728 <span class="k">push</span> <span class="i">@</span>{<span class="i">$FastaDataMap</span>{<span class="w">IDs</span>}}<span class="cm">,</span> <span class="i">$ID</span><span class="sc">;</span>
|
|
742 729 <span class="i">$FastaDataMap</span>{<span class="w">Description</span>}{<span class="i">$ID</span>} = <span class="i">$Description</span><span class="sc">;</span>
|
|
743 730 <span class="i">$FastaDataMap</span>{<span class="w">Count</span>} += <span class="n">1</span><span class="sc">;</span>
|
|
744 731 <span class="k">next</span> <span class="j">LINE</span><span class="sc">;</span>
|
|
745 732 <span class="s">}</span>
|
|
746 733 <span class="k">if</span> <span class="s">(</span><span class="i">$IgnoreID</span><span class="s">)</span> <span class="s">{</span> <span class="k">next</span> <span class="j">LINE</span><span class="sc">;</span> <span class="s">}</span>
|
|
747 734
|
|
748 735 <span class="c"># Remove any spaces in the sequence...</span>
|
|
749 736 <span class="i">$Line</span> =~ <span class="q">s/ //g</span><span class="sc">;</span>
|
|
750 737 <span class="c"># Sequence data for active ID...</span>
|
|
751 738 <span class="k">if</span> <span class="s">(</span><span class="k">exists</span> <span class="i">$FastaDataMap</span>{<span class="w">Sequence</span>}{<span class="i">$ID</span>}<span class="s">)</span> <span class="s">{</span>
|
|
752 739 <span class="i">$FastaDataMap</span>{<span class="w">Sequence</span>}{<span class="i">$ID</span>} .= <span class="i">$Line</span><span class="sc">;</span>
|
|
753 740 <span class="s">}</span>
|
|
754 741 <span class="k">else</span> <span class="s">{</span>
|
|
755 742 <span class="i">$FastaDataMap</span>{<span class="w">Sequence</span>}{<span class="i">$ID</span>} = <span class="i">$Line</span><span class="sc">;</span>
|
|
756 743 <span class="s">}</span>
|
|
757 744 <span class="s">}</span>
|
|
758 745 <span class="k">close</span> <span class="w">FASTAFILE</span><span class="sc">;</span>
|
|
759 746 <span class="k">return</span> \<span class="i">%FastaDataMap</span><span class="sc">;</span>
|
|
760 747 <span class="s">}</span>
|
|
761 748
|
|
762 749 <span class="c"># Read PIR fasta file and return a reference to hash with these keys:</span>
|
|
763 750 <span class="c">#</span>
|
|
764 751 <span class="c"># {IDs} - Array of sequence IDs</span>
|
|
765 752 <span class="c"># {Count} - Number of sequences</span>
|
|
766 753 <span class="c"># {Description}{$ID} - Sequence description</span>
|
|
767 754 <span class="c"># {Sequence}{$ID} - Sequence for a specific ID</span>
|
|
768 755 <span class="c"># {InputFileType} - Sequence file format</span>
|
|
769 756 <span class="c"># {ConservedAnnotation} - Conserved residue annonation</span>
|
|
770 757 <span class="c">#</span>
|
|
771 758 <span class="c"># Format:</span>
|
|
772 759 <span class="c"># A sequence in PIR format consists of:</span>
|
|
773 760 <span class="c"># One line starting with</span>
|
|
774 761 <span class="c"># a ">" (greater-than) sign, followed by</span>
|
|
775 762 <span class="c"># a two-letter code describing the sequence type code (P1, F1, DL, DC, RL, RC, N3, N1 or XX), followed by</span>
|
|
776 763 <span class="c"># a semicolon, followed by</span>
|
|
777 764 <span class="c"># the sequence identification code (the database ID-code).</span>
|
|
778 765 <span class="c"># One line containing a textual description of the sequence.</span>
|
|
779 766 <span class="c"># One or more lines containing the sequence itself. The end of the</span>
|
|
780 767 <span class="c"># sequence is marked by a "*" (asterisk) character.</span>
|
|
781 768 <span class="c">#</span>
|
|
782 769 <span class="c"># A file in PIR format may comprise more than one sequence.</span>
|
|
783 770 <span class="c">#</span>
|
|
784 771 <span class="c"># The PIR format is also often referred to as the NBRF format.</span>
|
|
785 772 <span class="c">#</span>
|
|
786 773 <span class="c"># Code SequenceType</span>
|
|
787 774 <span class="c"># P1 Protein (complete)</span>
|
|
788 775 <span class="c"># F1 Protein (fragment)</span>
|
|
789 776 <span class="c"># DL DNA (linear)</span>
|
|
790 777 <span class="c"># DC DNA (circular)</span>
|
|
791 778 <span class="c"># RL RNA (linear)</span>
|
|
792 779 <span class="c"># RC RNA (circular)</span>
|
|
793 780 <span class="c"># N3 tRNA</span>
|
|
794 781 <span class="c"># N1 Other functional RNA</span>
|
|
795 782 <span class="c">#</span>
|
|
796 783
|
|
797 <a name="_ReadPIRFastaFile-"></a> 784 <span class="k">sub </span><span class="m">_ReadPIRFastaFile</span> <span class="s">{</span>
|
|
798 785 <span class="k">my</span><span class="s">(</span><span class="i">$FastaFileName</span><span class="cm">,</span> <span class="i">$ID</span><span class="cm">,</span> <span class="i">$Description</span><span class="cm">,</span> <span class="i">$Line</span><span class="cm">,</span> <span class="i">$SequenceTypeCode</span><span class="cm">,</span> <span class="i">$ReadingSequenceData</span><span class="cm">,</span> <span class="i">%FastaDataMap</span><span class="s">)</span><span class="sc">;</span>
|
|
799 786
|
|
800 787 <span class="s">(</span><span class="i">$FastaFileName</span><span class="s">)</span> = <span class="i">@_</span><span class="sc">;</span>
|
|
801 788
|
|
802 789 <span class="i">%FastaDataMap</span> = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
803 790 <span class="i">@</span>{<span class="i">$FastaDataMap</span>{<span class="w">IDs</span>}} =<span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
804 791 <span class="i">%</span>{<span class="i">$FastaDataMap</span>{<span class="w">Description</span>}} =<span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
805 792 <span class="i">%</span>{<span class="i">$FastaDataMap</span>{<span class="w">Sequence</span>}} =<span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
806 793 <span class="i">%</span>{<span class="i">$FastaDataMap</span>{<span class="w">SequenceTypeCode</span>}} =<span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
807 794 <span class="i">$FastaDataMap</span>{<span class="w">Count</span>} = <span class="n">0</span><span class="sc">;</span>
|
|
808 795 <span class="i">$FastaDataMap</span>{<span class="w">InputFileType</span>} = <span class="q">'PIR'</span><span class="sc">;</span>
|
|
809 796
|
|
810 797 <span class="k">open</span> <span class="w">FASTAFILE</span><span class="cm">,</span> <span class="q">"$FastaFileName"</span> <span class="k">or</span> <span class="k">die</span> <span class="q">"Couldn't open $FastaFileName: $!\n"</span><span class="sc">;</span>
|
|
811 798 <span class="i">$ID</span> = <span class="q">''</span><span class="sc">;</span>
|
|
812 799 <span class="i">$ReadingSequenceData</span> = <span class="n">0</span><span class="sc">;</span>
|
|
813 800 <span class="j">LINE:</span> <span class="k">while</span> <span class="s">(</span><span class="i">$Line</span> = <span class="i">GetTextLine</span><span class="s">(</span>\<span class="i">*FASTAFILE</span><span class="s">)</span><span class="s">)</span> <span class="s">{</span>
|
|
814 801 <span class="k">if</span> <span class="s">(</span><span class="i">$Line</span> =~ <span class="q">/^\>/</span><span class="s">)</span> <span class="s">{</span>
|
|
815 802 <span class="c"># Start of a new ID...</span>
|
|
816 803 <span class="i">$Line</span> =~ <span class="q">s/^\>//</span><span class="sc">;</span>
|
|
817 804 <span class="i">$Line</span> = <span class="i">RemoveLeadingWhiteSpaces</span><span class="s">(</span><span class="i">$Line</span><span class="s">)</span><span class="sc">;</span>
|
|
818 805 <span class="s">(</span><span class="i">$SequenceTypeCode</span><span class="cm">,</span> <span class="i">$ID</span><span class="s">)</span> = <span class="q">/^\>(.*?)\;(.*?)$/</span><span class="sc">;</span>
|
|
819 806
|
|
820 807 <span class="c"># Use next line to retrieve sequence description...</span>
|
|
821 808 <span class="i">$Line</span> = <span class="i">GetTextLine</span><span class="s">(</span>\<span class="i">*FASTAFILE</span><span class="s">)</span><span class="sc">;</span>
|
|
822 809 <span class="i">$Line</span> = <span class="i">RemoveLeadingWhiteSpaces</span><span class="s">(</span><span class="i">$Line</span><span class="s">)</span><span class="sc">;</span>
|
|
823 810 <span class="i">$Description</span> = <span class="i">$Line</span><span class="sc">;</span>
|
|
824 811
|
|
825 812 <span class="k">if</span> <span class="s">(</span><span class="k">exists</span> <span class="i">$FastaDataMap</span>{<span class="w">Sequence</span>}{<span class="i">$ID</span>}<span class="s">)</span> <span class="s">{</span>
|
|
826 813 <span class="k">warn</span> <span class="q">"Warning: ID, $ID, in Fasta file already exists. Ignoring ID and sequence data...\n"</span><span class="sc">;</span>
|
|
827 814 <span class="k">next</span> <span class="j">LINE</span><span class="sc">;</span>
|
|
828 815 <span class="s">}</span>
|
|
829 816 <span class="i">$ReadingSequenceData</span> = <span class="n">1</span><span class="sc">;</span>
|
|
830 817 <span class="k">push</span> <span class="i">@</span>{<span class="i">$FastaDataMap</span>{<span class="w">IDs</span>}}<span class="cm">,</span> <span class="i">$ID</span><span class="sc">;</span>
|
|
831 818 <span class="i">$FastaDataMap</span>{<span class="w">SequenceTypeCode</span>}{<span class="i">$ID</span>} = <span class="i">$SequenceTypeCode</span><span class="sc">;</span>
|
|
832 819 <span class="i">$FastaDataMap</span>{<span class="w">Description</span>}{<span class="i">$ID</span>} = <span class="i">$Description</span><span class="sc">;</span>
|
|
833 820 <span class="i">$FastaDataMap</span>{<span class="w">Count</span>} += <span class="n">1</span><span class="sc">;</span>
|
|
834 821 <span class="k">next</span> <span class="j">LINE</span><span class="sc">;</span>
|
|
835 822 <span class="s">}</span>
|
|
836 823 <span class="k">if</span> <span class="s">(</span>!<span class="i">$ReadingSequenceData</span><span class="s">)</span> <span class="s">{</span> <span class="k">next</span> <span class="j">LINE</span><span class="sc">;</span> <span class="s">}</span>
|
|
837 824
|
|
838 825 <span class="c"># Remove any spaces in the sequence...</span>
|
|
839 826 <span class="i">$Line</span> =~ <span class="q">s/ //g</span><span class="sc">;</span>
|
|
840 827 <span class="k">if</span> <span class="s">(</span><span class="i">$Line</span> =~ <span class="q">/[\*]$/</span><span class="s">)</span> <span class="s">{</span>
|
|
841 828 <span class="c"># End of sequence...</span>
|
|
842 829 <span class="i">$ReadingSequenceData</span> = <span class="n">0</span><span class="sc">;</span>
|
|
843 830 <span class="i">$Line</span> =~ <span class="q">s/[\*]$//</span><span class="sc">;</span>
|
|
844 831 <span class="s">}</span>
|
|
845 832 <span class="c"># Sequence data for active ID...</span>
|
|
846 833 <span class="k">if</span> <span class="s">(</span><span class="k">exists</span> <span class="i">$FastaDataMap</span>{<span class="w">Sequence</span>}{<span class="i">$ID</span>}<span class="s">)</span> <span class="s">{</span>
|
|
847 834 <span class="i">$FastaDataMap</span>{<span class="w">Sequence</span>}{<span class="i">$ID</span>} .= <span class="i">$Line</span><span class="sc">;</span>
|
|
848 835 <span class="s">}</span>
|
|
849 836 <span class="k">else</span> <span class="s">{</span>
|
|
850 837 <span class="i">$FastaDataMap</span>{<span class="w">Sequence</span>}{<span class="i">$ID</span>} = <span class="i">$Line</span><span class="sc">;</span>
|
|
851 838 <span class="s">}</span>
|
|
852 839 <span class="s">}</span>
|
|
853 840 <span class="k">close</span> <span class="w">FASTAFILE</span><span class="sc">;</span>
|
|
854 841 <span class="k">return</span> \<span class="i">%FastaDataMap</span><span class="sc">;</span>
|
|
855 842 <span class="s">}</span>
|
|
856 843
|
|
857 844 <span class="c"># Read MSF file and return a reference to hash with these keys:</span>
|
|
858 845 <span class="c">#</span>
|
|
859 846 <span class="c"># {IDs} : Array of IDs in order as they appear in file</span>
|
|
860 847 <span class="c"># {Count}: ID count...</span>
|
|
861 848 <span class="c"># {Description}{$ID} : Description data...</span>
|
|
862 849 <span class="c"># {Sequence}{$ID} : Sequence data...</span>
|
|
863 850 <span class="c">#</span>
|
|
864 <a name="_ReadMSFFile-"></a> 851 <span class="k">sub </span><span class="m">_ReadMSFFile</span> <span class="s">{</span>
|
|
865 852 <span class="k">my</span><span class="s">(</span><span class="i">$MSFFileName</span><span class="cm">,</span> <span class="i">$Line</span><span class="cm">,</span> <span class="i">@LineWords</span><span class="cm">,</span> <span class="i">%MSFDataMap</span><span class="s">)</span><span class="sc">;</span>
|
|
866 853
|
|
867 854 <span class="s">(</span><span class="i">$MSFFileName</span><span class="s">)</span> = <span class="i">@_</span><span class="sc">;</span>
|
|
868 855
|
|
869 856 <span class="i">%MSFDataMap</span> = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
870 857 <span class="i">@</span>{<span class="i">$MSFDataMap</span>{<span class="w">IDs</span>}} =<span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
871 858 <span class="i">%</span>{<span class="i">$MSFDataMap</span>{<span class="w">Description</span>}} =<span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
872 859 <span class="i">%</span>{<span class="i">$MSFDataMap</span>{<span class="w">Sequence</span>}} =<span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
873 860 <span class="i">$MSFDataMap</span>{<span class="w">Count</span>} = <span class="n">0</span><span class="sc">;</span>
|
|
874 861 <span class="i">$MSFDataMap</span>{<span class="w">InputFileType</span>} = <span class="q">'MSF'</span><span class="sc">;</span>
|
|
875 862
|
|
876 863 <span class="k">open</span> <span class="w">MSFFILE</span><span class="cm">,</span> <span class="q">"$MSFFileName"</span> <span class="k">or</span> <span class="k">die</span> <span class="q">"Couldn't open $MSFFileName: $!\n"</span><span class="sc">;</span>
|
|
877 864
|
|
878 865 <span class="c"># Collect sequences and IDs...</span>
|
|
879 866 <span class="c">#</span>
|
|
880 867 <span class="c"># '//' after the name fields indicates end of header list and start of sequence data.</span>
|
|
881 868 <span class="c">#</span>
|
|
882 869 <span class="k">my</span><span class="s">(</span><span class="i">$ID</span><span class="cm">,</span> <span class="i">$Len</span><span class="cm">,</span> <span class="i">$Check</span><span class="cm">,</span> <span class="i">$Weight</span><span class="cm">,</span> <span class="i">$Sequence</span><span class="cm">,</span> <span class="i">$NameFieldsFound</span><span class="cm">,</span> <span class="i">%MSFIDsMap</span><span class="s">)</span><span class="sc">;</span>
|
|
883 870 <span class="i">%MSFIDsMap</span> = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
884 871 <span class="i">$NameFieldsFound</span> = <span class="n">0</span><span class="sc">;</span>
|
|
885 872 <span class="j">LINE:</span> <span class="k">while</span> <span class="s">(</span><span class="i">$Line</span> = <span class="i">GetTextLine</span><span class="s">(</span>\<span class="i">*MSFFILE</span><span class="s">)</span><span class="s">)</span> <span class="s">{</span>
|
|
886 873 <span class="k">if</span> <span class="s">(</span><span class="i">$Line</span> =~ <span class="q">/Name:/</span><span class="s">)</span> <span class="s">{</span>
|
|
887 874 <span class="i">$NameFieldsFound</span>++<span class="sc">;</span>
|
|
888 875 <span class="s">(</span><span class="i">$ID</span><span class="cm">,</span> <span class="i">$Len</span><span class="cm">,</span> <span class="i">$Check</span><span class="cm">,</span> <span class="i">$Weight</span><span class="s">)</span> = <span class="i">$Line</span> =~ <span class="q">/^[ ]*Name:[ ]+(.*?)[ ]+Len:[ ]+(.*?)[ ]+Check:[ ]+(.*?)[ ]+Weight:[ ]+(.*?)[ ]*$/</span><span class="sc">;</span>
|
|
889 876 <span class="k">if</span> <span class="s">(</span><span class="i">$ID</span> =~ <span class="q">/ /</span><span class="s">)</span> <span class="s">{</span>
|
|
890 877 <span class="s">(</span><span class="i">$ID</span><span class="s">)</span> = <span class="i">$ID</span> =~ <span class="q">/^(.*?)[ ]+/</span>
|
|
891 878 <span class="s">}</span>
|
|
892 879 <span class="k">if</span> <span class="s">(</span><span class="k">exists</span> <span class="i">$MSFIDsMap</span>{<span class="i">$ID</span>}<span class="s">)</span> <span class="s">{</span>
|
|
893 880 <span class="k">warn</span> <span class="q">"Warning: ID, $ID, in MSF file already exists. Ignoring ID and sequence data...\n"</span><span class="sc">;</span>
|
|
894 881 <span class="k">next</span> <span class="j">LINE</span><span class="sc">;</span>
|
|
895 882 <span class="s">}</span>
|
|
896 883 <span class="i">$MSFIDsMap</span>{<span class="i">$ID</span>} = <span class="i">$ID</span><span class="sc">;</span>
|
|
897 884 <span class="k">push</span> <span class="i">@</span>{<span class="i">$MSFDataMap</span>{<span class="w">IDs</span>}}<span class="cm">,</span> <span class="i">$ID</span><span class="sc">;</span>
|
|
898 885 <span class="i">$MSFDataMap</span>{<span class="w">Description</span>}{<span class="i">$ID</span>} = <span class="i">$ID</span><span class="sc">;</span>
|
|
899 886 <span class="i">$MSFDataMap</span>{<span class="w">Count</span>} += <span class="n">1</span><span class="sc">;</span>
|
|
900 887 <span class="s">}</span>
|
|
901 888 <span class="k">elsif</span> <span class="s">(</span> <span class="q">/\/\//</span> && <span class="i">$NameFieldsFound</span><span class="s">)</span> <span class="s">{</span>
|
|
902 889 <span class="c"># End of header list...</span>
|
|
903 890 <span class="k">last</span> <span class="j">LINE</span><span class="sc">;</span>
|
|
904 891 <span class="s">}</span>
|
|
905 892 <span class="s">}</span>
|
|
906 893 <span class="c"># Collect all sequences...</span>
|
|
907 894 <span class="c">#</span>
|
|
908 895 <span class="k">my</span><span class="s">(</span><span class="i">$FirstField</span><span class="cm">,</span> <span class="i">$SecondField</span><span class="s">)</span><span class="sc">;</span>
|
|
909 896 <span class="k">while</span> <span class="s">(</span><span class="i">$Line</span> = <span class="i">GetTextLine</span><span class="s">(</span>\<span class="i">*MSFFILE</span><span class="s">)</span><span class="s">)</span> <span class="s">{</span>
|
|
910 897 <span class="s">(</span><span class="i">$FirstField</span><span class="cm">,</span> <span class="i">$SecondField</span><span class="s">)</span> = <span class="i">$Line</span> =~ <span class="q">/^[ ]*(.*?)[ ]+(.*?)$/</span><span class="sc">;</span>
|
|
911 898 <span class="k">if</span> <span class="s">(</span><span class="k">exists</span> <span class="i">$MSFIDsMap</span>{<span class="i">$FirstField</span>}<span class="s">)</span> <span class="s">{</span>
|
|
912 899 <span class="c"># It's ID and sequence data...</span>
|
|
913 900 <span class="i">$ID</span> = <span class="i">$FirstField</span><span class="sc">;</span>
|
|
914 901 <span class="i">$Sequence</span> = <span class="i">$SecondField</span><span class="sc">;</span>
|
|
915 902 <span class="c"># Take out spaces and leave the gap characters...</span>
|
|
916 903 <span class="i">$Sequence</span> =~ <span class="q">s/ //g</span><span class="sc">;</span>
|
|
917 904 <span class="k">if</span> <span class="s">(</span><span class="i">$MSFDataMap</span>{<span class="w">Sequence</span>}{<span class="i">$ID</span>}<span class="s">)</span> <span class="s">{</span>
|
|
918 905 <span class="i">$MSFDataMap</span>{<span class="w">Sequence</span>}{<span class="i">$ID</span>} .= <span class="i">$Sequence</span><span class="sc">;</span>
|
|
919 906 <span class="s">}</span>
|
|
920 907 <span class="k">else</span> <span class="s">{</span>
|
|
921 908 <span class="i">$MSFDataMap</span>{<span class="w">Sequence</span>}{<span class="i">$ID</span>} = <span class="i">$Sequence</span><span class="sc">;</span>
|
|
922 909 <span class="s">}</span>
|
|
923 910 <span class="s">}</span>
|
|
924 911 <span class="s">}</span>
|
|
925 912
|
|
926 913 <span class="k">close</span> <span class="w">MSFFILE</span><span class="sc">;</span>
|
|
927 914 <span class="k">return</span> \<span class="i">%MSFDataMap</span><span class="sc">;</span>
|
|
928 915 <span class="s">}</span>
|
|
929 916
|
|
930 917
|
|
931 <a name="EOF-"></a></pre>
|
|
932 <p> </p>
|
|
933 <br />
|
|
934 <center>
|
|
935 <img src="../../../images/h2o2.png">
|
|
936 </center>
|
|
937 </body>
|
|
938 </html>
|