annotate variant_effect_predictor/Bio/AlignIO/pfam.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: pfam.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::pfam
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
4
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
5 # based on the Bio::SeqIO:: modules
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::pfam - pfam 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 Bio::SimpleAlign objects to and from pfam 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::pfam;
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
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
60 use Bio::AlignIO;
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
61 use Bio::SimpleAlign;
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
62 @ISA = qw(Bio::AlignIO);
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
63
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
64 =head2 next_aln
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
65
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
66 Title : next_aln
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
67 Usage : $aln = $stream->next_aln()
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
68 Function: returns the next alignment in the stream
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
69 Returns : L<Bio::Align::AlignI> object
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
70 Args : NONE
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
71
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
72 =cut
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
73
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
74 sub next_aln {
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
75 my $self = shift;
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
76 my $entry;
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
77 my $name;
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
78 my $start;
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
79 my $end;
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
80 my $seq;
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
81 my $add;
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
82 my $acc;
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
83 my %names;
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
84
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
85 my $aln = Bio::SimpleAlign->new(-source => 'pfam');
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 chomp $entry;
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
89 $entry =~ /^\/\// && last;
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
90 if($entry !~ /^(\S+)\/(\d+)-(\d+)\s+(\S+)\s*/ ) {
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
91 $self->throw("Found a bad line [$_] in the pfam format alignment");
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
92 next;
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
93 }
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
94
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
95 $name = $1;
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
96 $start = $2;
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
97 $end = $3;
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
98 $seq = $4;
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
99
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
100
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
101 $add = new Bio::LocatableSeq('-seq'=>$seq,
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
102 '-id'=>$name,
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
103 '-start'=>$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 $aln->add_seq($add);
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
108
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
109 }
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
110
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
111 # If $end <= 0, we have either reached the end of
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
112 # file in <> or we have encountered some other error
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
113 #
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
114 if ($end <= 0) { undef $aln;}
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
115
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
116 return $aln;
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
117 }
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
118
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
119
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
120
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
121 =head2 write_aln
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
122
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
123 Title : write_aln
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
124 Usage : $stream->write_aln(@aln)
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
125 Function: writes the $aln object into the stream
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
126 Returns : 1 for success and 0 for error
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
127 Args : L<Bio::Align::AlignI> object
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
128
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
129
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
130 =cut
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
131
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
132 sub write_aln {
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
133 my ($self,@aln) = @_;
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
134 if( @aln > 1 ) { $self->warn("Only the 1st pfam alignment will be output since the format does not support multiple alignments in the same file"); }
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
135 my $aln = shift @aln;
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
136 if( ! $aln || ! $aln->isa('Bio::Align::AlignI') ) {
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
137 $self->warn("Must provide a Bio::Align::AlignI object when calling write_aln");
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
138 next;
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
139 }
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
140 my ($namestr,$seq,$add);
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
141 my ($maxn);
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
142 $maxn = $aln->maxdisplayname_length();
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
143
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
144 foreach $seq ( $aln->each_seq() ) {
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
145 $namestr = $aln->displayname($seq->get_nse());
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
146 $add = $maxn - length($namestr) + 2;
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
147 $namestr .= " " x $add;
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
148 $self->_print (sprintf("%s %s\n",$namestr,$seq->seq())) or return;
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
149 }
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
150 $self->flush() if $self->_flush_on_write && defined $self->_fh;
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
151 return 1;
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
152 }
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
153
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
154 1;