annotate variant_effect_predictor/Bio/AlignIO/psi.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: psi.pm,v 1.6 2002/12/23 19:36:39 jason Exp $
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
2 #
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
3 # BioPerl module for Bio::AlignIO::psi
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
4 #
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
5 # Cared for by Jason Stajich <jason@bioperl.org>
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
6 #
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
7 # Copyright Jason Stajich
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
8 #
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
9 # You may distribute this module under the same terms as perl itself
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
10
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
11 # POD documentation - main docs before the code
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
12
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
13 =head1 NAME
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
14
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
15 Bio::AlignIO::psi - Read/Write PSI-BLAST profile alignment files
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
16
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
17 =head1 SYNOPSIS
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
18
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
19 This module will parse PSI-BLAST output of the format seqid XXXX
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
20
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
21 =head1 DESCRIPTION
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
22
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
23 Describe the object here
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
24
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
25 =head1 FEEDBACK
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
26
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
27 =head2 Mailing Lists
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
28
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
29 User feedback is an integral part of the evolution of this and other
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
30 Bioperl modules. Send your comments and suggestions preferably to
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
31 the Bioperl mailing list. Your participation is much appreciated.
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
32
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
33 bioperl-l@bioperl.org - General discussion
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
34 http://bioperl.org/MailList.shtml - About the mailing lists
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
35
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
36 =head2 Reporting Bugs
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
37
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
38 Report bugs to the Bioperl bug tracking system to help us keep track
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
39 of the bugs and their resolution. Bug reports can be submitted via
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
40 email or the web:
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
41
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
42 bioperl-bugs@bioperl.org
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
43 http://bugzilla.bioperl.org/
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
44
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
45 =head1 AUTHOR - Jason Stajich
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
46
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
47 Email jason@bioperl.org
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
48
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
49 Describe contact details here
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
50
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
51 =head1 CONTRIBUTORS
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
52
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
53 Additional contributors names and emails here
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
54
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
55 =head1 APPENDIX
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
56
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
57 The rest of the documentation details each of the object methods.
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
58 Internal methods are usually preceded with a _
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
59
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
60 =cut
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
61
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
62
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
63 # Let the code begin...
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
64
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
65
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
66 package Bio::AlignIO::psi;
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
67 use vars qw(@ISA $BlockLen $IdLength);
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
68 use strict;
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
69
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
70 $BlockLen = 100;
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
71 $IdLength = 13;
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
72
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
73 # Object preamble - inherits from Bio::Root::Root
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
74
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
75 use Bio::SimpleAlign;
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
76 use Bio::AlignIO;
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
77 use Bio::LocatableSeq;
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
78
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
79 @ISA = qw(Bio::AlignIO);
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
80
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
81 =head2 new
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
82
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
83 Title : new
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
84 Usage : my $obj = new Bio::AlignIO::psi();
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
85 Function: Builds a new Bio::AlignIO::psi object
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
86 Returns : Bio::AlignIO::psi
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
87 Args :
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
88
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
89 =cut
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
90
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
91 =head2 next_aln
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
92
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
93 Title : next_aln
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
94 Usage : $aln = $stream->next_aln()
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
95 Function: returns the next alignment in the stream
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
96 Returns : L<Bio::Align::AlignI> object
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
97 Args : NONE
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
98
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
99 =cut
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
100
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
101 sub next_aln {
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
102 my ($self) = @_;
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
103 my $aln;
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
104 my %seqs;
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
105 my @order;
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
106 while( defined ($_ = $self->_readline ) ) {
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
107 next if( /^\s+$/);
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
108 if( !defined $aln ) {
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
109 $aln = new Bio::SimpleAlign;
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
110 }
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
111 my ($id,$s) = split;
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
112 push @order, $id if( ! defined $seqs{$id});
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
113 $seqs{$id} .= $s;
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
114 }
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
115 foreach my $id ( @order) {
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
116 my $seq = new Bio::LocatableSeq(-seq => $seqs{$id},
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
117 -id => $id,
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
118 -start => 1,
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
119 -end => length($seqs{$id}));
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
120 $aln->add_seq($seq);
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
121 }
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
122 return $aln;
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
123 }
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
124
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
125 =head2 write_aln
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
126
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
127 Title : write_aln
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
128 Usage : $stream->write_aln(@aln)
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
129 Function: writes the NCBI psi-format object (.aln) into the stream
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
130 Returns : 1 for success and 0 for error
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
131 Args : L<Bio::Align::AlignI> object
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
132
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
133
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
134 =cut
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
135
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
136 sub write_aln {
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
137 my ($self,$aln) = @_;
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
138 unless( defined $aln && ref($aln) &&
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
139 $aln->isa('Bio::Align::AlignI') ) {
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
140 $self->warn("Must provide a valid Bio::Align::AlignI to write_aln");
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
141 return 0;
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
142 }
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
143 my $ct = 0;
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
144 my @seqs = $aln->each_seq;
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
145 my $len = 1;
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
146 my $alnlen = $aln->length;
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
147 my $idlen = $IdLength;
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
148 my @ids = map { substr($_->display_id,0,$idlen) } @seqs;
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
149 while( $len < $alnlen ) {
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
150 my $start = $len;
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
151 my $end = $len + $BlockLen;
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
152 if( $end > $alnlen ) { $end = $alnlen; }
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
153 my $c = 0;
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
154 foreach my $seq ( @seqs ) {
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
155 $self->_print(sprintf("%-".$idlen."s %s\n",
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
156 $ids[$c++],
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
157 $seq->subseq($start,$end)));
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
158 }
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
159 $self->_print("\n");
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
160 $len += $BlockLen+1;
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
161 }
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
162 $self->flush if $self->_flush_on_write && defined $self->_fh;
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
163 return 1;
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
164 }
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
165
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
166 1;