annotate variant_effect_predictor/Bio/AlignIO/mase.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: mase.pm,v 1.9 2002/10/22 07:38:25 lapp Exp $
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
2 #
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
3 # BioPerl module for Bio::AlignIO::mase
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
4
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
5 # based on the Bio::SeqIO::mase 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::mase - mase 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 mase 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::mase;
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
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
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
65 =head2 next_aln
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
66
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
67 Title : next_aln
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
68 Usage : $aln = $stream->next_aln()
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
69 Function: returns the next alignment in the stream.
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
70 Returns : L<Bio::Align::AlignI> object
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
71 Args : NONE
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
72
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
73 =cut
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
74
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
75 sub next_aln {
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
76 my $self = shift;
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
77 my $entry;
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
78 my $name;
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
79 my $start;
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
80 my $end;
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
81 my $seq;
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
82 my $add;
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
83 my $count = 0;
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
84 my $seq_residues;
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
85
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
86 my $aln = Bio::SimpleAlign->new(-source => 'mase');
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
87
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
88
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
89 while( $entry = $self->_readline) {
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
90 $entry =~ /^;/ && next;
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
91 if( $entry =~ /^(\S+)\/(\d+)-(\d+)/ ) {
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
92 $name = $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 $entry =~ s/\s//g;
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
97 $name = $entry;
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
98 $end = -1;
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
99 }
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
100
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
101 $seq = "";
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
102
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
103 while( $entry = $self->_readline) {
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
104 $entry =~ /^;/ && last;
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
105 $entry =~ s/[^A-Za-z\.\-]//g;
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
106 $seq .= $entry;
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
107 }
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
108 if( $end == -1) {
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
109 $start = 1;
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
110
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
111 $seq_residues = $seq;
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
112 $seq_residues =~ s/\W//g;
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
113 $end = length($seq_residues);
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
114 }
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
115
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
116 $add = new Bio::LocatableSeq('-seq'=>$seq,
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
117 '-id'=>$name,
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
118 '-start'=>$start,
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
119 '-end'=>$end,
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
120 );
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
121
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
122
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
123 $aln->add_seq($add);
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
124
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
125
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
126 # If $end <= 0, we have either reached the end of
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
127 # file in <> or we have encountered some other error
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
128 #
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
129 if ($end <= 0) { undef $aln;}
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
130
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
131 }
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
132
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
133 return $aln;
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
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
138 =head2 write_aln
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
139
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
140 Title : write_aln
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
141 Usage : $stream->write_aln(@aln)
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
142 Function: writes the $aln object into the stream in mase format ###Not yet implemented!###
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
143 Returns : 1 for success and 0 for error
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
144 Args : L<Bio::Align::AlignI> object
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
145
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
146
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
147 =cut
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
148
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
149 sub write_aln {
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
150 my ($self,@aln) = @_;
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
151
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
152 $self->throw("Sorry: mase-format output, not yet implemented! /n");
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
153 }
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
154
2bc9b66ada89 Uploaded
mahtabm
parents:
diff changeset
155 1;