annotate variant_effect_predictor/Bio/AlignIO/selex.pm @ 0:2bc9b66ada89 draft default tip

Uploaded
author mahtabm
date Thu, 11 Apr 2013 06:29:17 -0400
parents
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
1 # $Id: selex.pm,v 1.10 2002/10/22 07:38:26 lapp Exp $
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
2 #
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
3 # BioPerl module for Bio::AlignIO::selex
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
4
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
5 # based on the Bio::SeqIO::selex module
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
6 # by Ewan Birney <birney@sanger.ac.uk>
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
7 # and Lincoln Stein <lstein@cshl.org>
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
8 #
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
9 # and the SimpleAlign.pm module of Ewan Birney
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
10 #
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
11 # Copyright Peter Schattner
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
12 #
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
13 # You may distribute this module under the same terms as perl itself
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
14 # _history
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
15 # September 5, 2000
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
16 # POD documentation - main docs before the code
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
17
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
18 =head1 NAME
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
19
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
20 Bio::AlignIO::selex - selex sequence input/output stream
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
21
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
22 =head1 SYNOPSIS
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
23
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
24 Do not use this module directly. Use it via the L<Bio::AlignIO> class.
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
25
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
26 =head1 DESCRIPTION
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
27
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
28 This object can transform L<Bio::Align::AlignI> objects to and from selex flat
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
29 file databases.
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
30
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
31 =head1 FEEDBACK
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
32
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
33 =head2 Reporting Bugs
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
34
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
35 Report bugs to the Bioperl bug tracking system to help us keep track
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
36 the bugs and their resolution.
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
37 Bug reports can be submitted via email or the web:
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
38
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
39 bioperl-bugs@bio.perl.org
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
40 http://bugzilla.bioperl.org/
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
41
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
42 =head1 AUTHORS - Peter Schattner
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
43
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
44 Email: schattner@alum.mit.edu
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
45
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
46
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
47 =head1 APPENDIX
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
48
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
49 The rest of the documentation details each of the object
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
50 methods. Internal methods are usually preceded with a _
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
51
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
52 =cut
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
53
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
54 # Let the code begin...
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
55
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
56 package Bio::AlignIO::selex;
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
57 use vars qw(@ISA);
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
58 use strict;
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
59 use Bio::AlignIO;
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
60
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
61 @ISA = qw(Bio::AlignIO);
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
62
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
63 =head2 next_aln
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
64
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
65 Title : next_aln
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
66 Usage : $aln = $stream->next_aln()
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
67 Function: returns the next alignment in the stream. Tries to read *all* selex
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
68 It reads all non whitespace characters in the alignment
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
69 area. For selexs with weird gaps (eg ~~~) map them by using
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
70 $al->map_chars('~','-')
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
71 Returns : L<Bio::Align::AlignI> object
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
72 Args : NONE
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
73
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
74 =cut
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
75
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
76 sub next_aln {
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
77 my $self = shift;
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
78 my $entry;
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
79 my ($start,$end,%align,$name,$seqname,$seq,$count,%hash,%c2name, %accession, $no);
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
80 my $aln = Bio::SimpleAlign->new(-source => 'selex');
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
81
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
82 # in selex format, every non-blank line that does not start
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
83 # with '#=' is an alignment segment; the '#=' lines are mark up lines.
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
84 # Of particular interest are the '#=GF <name/st-ed> AC <accession>'
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
85 # lines, which give accession numbers for each segment
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
86
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
87 while( $entry = $self->_readline) {
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
88 $entry =~ /^\#=GS\s+(\S+)\s+AC\s+(\S+)/ && do {
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
89 $accession{ $1 } = $2;
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
90 next;
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
91 };
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
92 $entry !~ /^([^\#]\S+)\s+([A-Za-z\.\-]+)\s*/ && next;
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
93
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
94 $name = $1;
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
95 $seq = $2;
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
96
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
97 if( ! defined $align{$name} ) {
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
98 $count++;
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
99 $c2name{$count} = $name;
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
100 }
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
101 $align{$name} .= $seq;
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
102 }
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
103
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
104 # ok... now we can make the sequences
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
105
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
106 $count = 0;
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
107 foreach $no ( sort { $a <=> $b } keys %c2name ) {
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
108 $name = $c2name{$no};
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
109
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
110 if( $name =~ /(\S+)\/(\d+)-(\d+)/ ) {
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
111 $seqname = $1;
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
112 $start = $2;
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
113 $end = $3;
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
114 } else {
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
115 $seqname=$name;
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
116 $start = 1;
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
117 $end = length($align{$name});
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
118 }
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
119 $seq = new Bio::LocatableSeq('-seq'=>$align{$name},
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
120 '-id'=>$seqname,
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
121 '-start'=>$start,
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
122 '-end'=>$end,
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
123 '-type'=>'aligned',
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
124 '-accession_number' => $accession{$name},
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
125
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
126 );
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
127
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
128 $aln->add_seq($seq);
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
129 $count++;
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
130 }
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
131
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
132 # If $end <= 0, we have either reached the end of
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
133 # file in <> or we have encountered some other error
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
134 #
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
135 if ($end <= 0) { undef $aln;}
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
136
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
137 return $aln;
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
138 }
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
139
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
140
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
141 =head2 write_aln
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
142
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
143 Title : write_aln
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
144 Usage : $stream->write_aln(@aln)
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
145 Function: writes the $aln object into the stream in selex format
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
146 Returns : 1 for success and 0 for error
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
147 Args : L<Bio::Align::AlignI> object
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
148
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
149
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
150 =cut
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
151
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
152 sub write_aln {
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
153 my ($self,@aln) = @_;
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
154 my ($namestr,$seq,$add);
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
155 my ($maxn);
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
156 foreach my $aln (@aln) {
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
157 $maxn = $aln->maxdisplayname_length();
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
158 foreach $seq ( $aln->each_seq() ) {
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
159 $namestr = $aln->displayname($seq->get_nse());
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
160 $add = $maxn - length($namestr) + 2;
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
161 $namestr .= " " x $add;
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
162 $self->_print (sprintf("%s %s\n",$namestr,$seq->seq())) or return;
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
163 }
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
164 }
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
165 $self->flush if $self->_flush_on_write && defined $self->_fh;
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
166 return 1;
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
167 }
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
168
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
169 1;