annotate variant_effect_predictor/Bio/AlignIO/fasta.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: fasta.pm,v 1.11 2002/12/14 19:09:05 birney Exp $
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
2 #
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
3 # BioPerl module for Bio::AlignIO::fasta
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
4
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
5 # based on the Bio::SeqIO::fasta 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::fasta - FastA MSA 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::SimpleAlign> objects to and from
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
29 fasta flat file databases. This is for the fasta sequence format NOT
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
30 FastA analysis program. To process the pairwise alignments from a
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
31 FastA (FastX, FastN, FastP, tFastA, etc) use the Bio::SearchIO module.
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
32
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
33 =head1 FEEDBACK
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
34
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
35 =head2 Reporting Bugs
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
36
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
37 Report bugs to the Bioperl bug tracking system to help us keep track
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
38 the bugs and their resolution. Bug reports can be submitted via email
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
39 or the web:
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
40
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
41 bioperl-bugs@bio.perl.org
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
42 http://bugzilla.bioperl.org/
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
43
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
44 =head1 AUTHORS - Peter Schattner
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
45
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
46 Email: schattner@alum.mit.edu
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
47
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
48
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
49 =head1 APPENDIX
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
50
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
51 The rest of the documentation details each of the object
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
52 methods. Internal methods are usually preceded with a _
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
53
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
54 =cut
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
55
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
56 # Let the code begin...
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
57
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
58 package Bio::AlignIO::fasta;
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
59 use vars qw(@ISA);
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
60 use strict;
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
61
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
62 use Bio::AlignIO;
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
63 use Bio::SimpleAlign;
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
64
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
65 @ISA = qw(Bio::AlignIO);
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
66
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
67
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
68 =head2 next_aln
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
69
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
70 Title : next_aln
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
71 Usage : $aln = $stream->next_aln()
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
72 Function: returns the next alignment in the stream.
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
73 Returns : L<Bio::Align::AlignI> object - returns 0 on end of file
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
74 or on error
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
75 Args : NONE
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
76
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
77 =cut
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
78
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
79 sub next_aln {
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
80 my $self = shift;
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
81 my $entry;
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
82 my ($start,$end,$name,$seqname,$seq,$seqchar,$tempname,%align);
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
83 my $aln = Bio::SimpleAlign->new();
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
84
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
85 while(defined ($entry = $self->_readline)) {
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
86 if($entry =~ /^>(\S+)/ ) {
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
87 $tempname = $1;
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
88 if( defined $name ) {
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
89 # put away last name and sequence
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
90
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
91 if( $name =~ /(\S+)\/(\d+)-(\d+)/ ) {
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
92 $seqname = $1;
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
93 $start = $2;
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
94 $end = $3;
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
95 } else {
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
96 $seqname=$name;
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
97 $start = 1;
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
98 $end = length($seqchar); #ps 9/6/00
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
99 }
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
100 # print STDERR "Going to add with $seqchar $seqname\n";
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
101 $seq = new Bio::LocatableSeq('-seq'=>$seqchar,
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
102 '-id'=>$seqname,
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
103 '-start'=>$start,
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
104 '-end'=>$end,
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
105 );
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
106
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
107 $aln->add_seq($seq);
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
108 }
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
109 $name = $tempname;
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
110 $seqchar = "";
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
111 next;
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
112 }
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
113 $entry =~ s/[^A-Za-z\.\-]//g;
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
114 $seqchar .= $entry;
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
115
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
116 }
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
117 #
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
118 # Next two lines are to silence warnings that
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
119 # otherwise occur at EOF when using <$fh>
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
120
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
121 if (!defined $name) {$name="";}
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
122 if (!defined $seqchar) {$seqchar="";}
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
123
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
124 # Put away last name and sequence
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
125 if( $name =~ /(\S+)\/(\d+)-(\d+)/ ) {
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
126 $seqname = $1;
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
127 $start = $2;
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
128 $end = $3;
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
129 } else {
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
130 $seqname=$name;
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
131 $start = 1;
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
132 $end = length($seqchar); #ps 9/6/00
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
133 # $end = length($align{$name});
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
134 }
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
135
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
136
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
137 # If $end <= 0, we have either reached the end of
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
138 # file in <> or we have encountered some other error
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
139 #
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
140 if ($end <= 0) { undef $aln; return $aln;}
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
141
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
142 # This logic now also reads empty lines at the
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
143 # end of the file. Skip this is seqchar and seqname is null
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
144 if( length($seqchar) == 0 && length($seqname) == 0 ) {
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
145 # skip
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
146 } else {
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
147 # print STDERR "end to add with $seqchar $seqname\n";
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
148 $seq = new Bio::LocatableSeq('-seq'=>$seqchar,
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
149 '-id'=>$seqname,
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
150 '-start'=>$start,
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
151 '-end'=>$end,
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
152 );
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
153
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
154 $aln->add_seq($seq);
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
155 }
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
156
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
157 return $aln;
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
158
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
159 }
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
160
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
161
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
162 =head2 write_aln
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
163
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
164 Title : write_aln
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
165 Usage : $stream->write_aln(@aln)
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
166 Function: writes the $aln object into the stream in fasta format
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
167 Returns : 1 for success and 0 for error
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
168 Args : L<Bio::Align::AlignI> object
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
169
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
170
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
171 =cut
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
172
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
173 sub write_aln {
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
174 my ($self,@aln) = @_;
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
175 my ($seq,$rseq,$name,$count,$length,$seqsub);
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
176
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
177 foreach my $aln (@aln) {
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
178 if( ! $aln || ! $aln->isa('Bio::Align::AlignI') ) {
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
179 $self->warn("Must provide a Bio::Align::AlignI object when calling write_aln");
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
180 next;
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
181 }
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
182 foreach $rseq ( $aln->each_seq() ) {
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
183 $name = $aln->displayname($rseq->get_nse());
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
184 $seq = $rseq->seq();
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
185 $self->_print (">$name\n") or return ;
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
186 $count =0;
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
187 $length = length($seq);
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
188 while( ($count * 60 ) < $length ) {
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
189 $seqsub = substr($seq,$count*60,60);
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
190 $self->_print ("$seqsub\n") or return ;
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
191 $count++;
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
192 }
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
193 }
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
194 }
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
195 $self->flush if $self->_flush_on_write && defined $self->_fh;
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
196 return 1;
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
197 }
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
198
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
199 1;