annotate variant_effect_predictor/Bio/AlignIO/phylip.pm @ 1:d6778b5d8382 draft default tip

Deleted selected files
author willmclaren
date Fri, 03 Aug 2012 10:05:43 -0400
parents 21066c0abaf5
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
1 # $Id: phylip.pm,v 1.24.2.1 2003/01/26 15:52:30 jason Exp $
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
2 #
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
3 # BioPerl module for Bio::AlignIO::phylip
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
4 #
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
5 # Copyright Heikki Lehvaslaiho
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
6 #
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
7
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
8 =head1 NAME
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
9
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
10 Bio::AlignIO::phylip - PHYLIP format sequence input/output stream
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
11
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
12 =head1 SYNOPSIS
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
13
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
14 # Do not use this module directly. Use it via the Bio::AlignIO class.
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
15
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
16 use Bio::AlignIO;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
17 use Bio::SimpleAlign;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
18 #you can set the name length to something other than the default 10
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
19 #if you use a version of phylip (hacked) that accepts ids > 10
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
20 my $phylipstream = new Bio::AlignIO(-format => 'phylip',
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
21 -fh => \*STDOUT,
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
22 -idlength=>30);
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
23 # convert data from one format to another
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
24 my $gcgstream = new Bio::AlignIO(-format => 'msf',
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
25 -file => 't/data/cysprot1a.msf');
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
26
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
27 while( my $aln = $gcgstream->next_aln ) {
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
28 $phylipstream->write_aln($aln);
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
29 }
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
30
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
31 # do it again with phylip sequential format format
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
32 $phylipstream->interleaved(0);
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
33 # can also initialize the object like this
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
34 $phylipstream = new Bio::AlignIO(-interleaved => 0,
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
35 -format => 'phylip',
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
36 -fh => \*STDOUT,
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
37 -idlength=>10);
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
38 $gcgstream = new Bio::AlignIO(-format => 'msf',
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
39 -file => 't/data/cysprot1a.msf');
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
40
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
41 while( my $aln = $gcgstream->next_aln ) {
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
42 $phylipstream->write_aln($aln);
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
43 }
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
44
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
45 =head1 DESCRIPTION
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
46
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
47 This object can transform Bio::SimpleAlign objects to and from PHYLIP
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
48 interleaved format. It will not work with PHYLIP sequencial format.
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
49
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
50 This module will output PHYLIP sequential format. By specifying the
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
51 flag -interleaved =E<gt> 0 in the initialization the module can output
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
52 data in interleaved format.
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
53
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
54 =head1 FEEDBACK
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
55
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
56 =head2 Reporting Bugs
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
57
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
58 Report bugs to the Bioperl bug tracking system to help us keep track
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
59 the bugs and their resolution.
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
60 Bug reports can be submitted via email or the web:
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
61
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
62 bioperl-bugs@bio.perl.org
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
63 http://bugzilla.bioperl.org/
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
64
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
65 =head1 AUTHORS - Heikki Lehvaslaiho and Jason Stajich
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
66
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
67 Email: heikki@ebi.ac.uk
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
68 Email: jason@bioperl.org
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
69
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
70 =head1 APPENDIX
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
71
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
72 The rest of the documentation details each of the object
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
73 methods. Internal methods are usually preceded with a _
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
74
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
75 =cut
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
76
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
77 # Let the code begin...
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
78
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
79 package Bio::AlignIO::phylip;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
80 use vars qw(@ISA $DEFAULTIDLENGTH $DEFAULTLINELEN);
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
81 use strict;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
82
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
83 use Bio::SimpleAlign;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
84 use Bio::AlignIO;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
85
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
86 @ISA = qw(Bio::AlignIO);
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
87
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
88 BEGIN {
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
89 $DEFAULTIDLENGTH = 10;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
90 $DEFAULTLINELEN = 60;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
91 }
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
92
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
93 =head2 new
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
94
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
95 Title : new
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
96 Usage : my $alignio = new Bio::AlignIO(-format => 'phylip'
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
97 -file => '>file',
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
98 -idlength => 10,
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
99 -idlinebreak => 1);
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
100 Function: Initialize a new L<Bio::AlignIO::phylip> reader or writer
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
101 Returns : L<Bio::AlignIO> object
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
102 Args : [specific for writing of phylip format files]
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
103 -idlength => integer - length of the id (will pad w/
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
104 spaces if needed)
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
105 -interleaved => boolean - whether or not write as interleaved
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
106 or sequential format
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
107 -linelength => integer of how long a sequence lines should be
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
108 -idlinebreak => insert a line break after the sequence id
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
109 so that sequence starts on the next line
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
110
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
111 =cut
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
112
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
113 sub _initialize {
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
114 my($self,@args) = @_;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
115 $self->SUPER::_initialize(@args);
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
116
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
117 my ($interleave,$linelen,$idlinebreak,
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
118 $idlength) = $self->_rearrange([qw(INTERLEAVED
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
119 LINELENGTH
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
120 IDLINEBREAK
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
121 IDLENGTH)],@args);
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
122 $self->interleaved(1) if( $interleave || ! defined $interleave);
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
123 $self->idlength($idlength || $DEFAULTIDLENGTH);
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
124 $self->id_linebreak(1) if( $idlinebreak );
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
125 $self->line_length($linelen) if defined $linelen && $linelen > 0;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
126 1;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
127 }
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
128
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
129 =head2 next_aln
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
130
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
131 Title : next_aln
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
132 Usage : $aln = $stream->next_aln()
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
133 Function: returns the next alignment in the stream.
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
134 Throws an exception if trying to read in PHYLIP
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
135 sequential format.
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
136 Returns : L<Bio::SimpleAlign> object
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
137 Args :
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
138
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
139 =cut
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
140
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
141 sub next_aln {
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
142 my $self = shift;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
143 my $entry;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
144 my ($seqcount, $residuecount, %hash, $name,$str,
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
145 @names,$seqname,$start,$end,$count,$seq);
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
146
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
147 my $aln = Bio::SimpleAlign->new(-source => 'phylip');
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
148 $entry = $self->_readline and
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
149 ($seqcount, $residuecount) = $entry =~ /\s*(\d+)\s+(\d+)/;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
150 return 0 unless $seqcount and $residuecount;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
151
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
152 # first alignment section
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
153 my $idlen = $self->idlength;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
154 $count = 0;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
155 my $non_interleaved = ! $self->interleaved ;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
156 while( $entry = $self->_readline) {
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
157 last if( $entry =~ /^\s?$/ && ! $non_interleaved );
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
158
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
159 if( $entry =~ /^\s+(.+)$/ ) {
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
160 $str = $1;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
161 $non_interleaved = 1;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
162 $str =~ s/\s//g;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
163 $count = scalar @names;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
164 $hash{$count} .= $str;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
165 } elsif( $entry =~ /^(.{$idlen})\s+(.*)\s$/ ) {
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
166 $name = $1;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
167 $str = $2;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
168 $name =~ s/[\s\/]/_/g;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
169 $name =~ s/_+$//; # remove any trailing _'s
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
170 push @names, $name;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
171 $str =~ s/\s//g;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
172 $count = scalar @names;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
173 $hash{$count} = $str;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
174 }
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
175 $self->throw("Not a valid interleaved PHYLIP file!") if $count > $seqcount;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
176 }
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
177
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
178 unless( $non_interleaved ) {
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
179 # interleaved sections
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
180 $count = 0;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
181 while( $entry = $self->_readline) {
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
182 # finish current entry
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
183 if($entry =~/\s*\d+\s+\d+/){
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
184 $self->_pushback($entry);
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
185 last;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
186 }
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
187 $count = 0, next if $entry =~ /^\s$/;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
188
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
189 $entry =~ /\s*(.*)$/ && do {
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
190 $str = $1;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
191 $str =~ s/\s//g;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
192 $count++;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
193 $hash{$count} .= $str;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
194 };
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
195 $self->throw("Not a valid interleaved PHYLIP file!") if $count > $seqcount;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
196 }
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
197 }
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
198 return 0 if scalar @names < 1;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
199
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
200 # sequence creation
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
201 $count = 0;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
202 foreach $name ( @names ) {
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
203 $count++;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
204 if( $name =~ /(\S+)\/(\d+)-(\d+)/ ) {
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
205 $seqname = $1;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
206 $start = $2;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
207 $end = $3;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
208 } else {
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
209 $seqname=$name;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
210 $start = 1;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
211 $str = $hash{$count};
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
212 $str =~ s/[^A-Za-z]//g;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
213 $end = length($str);
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
214 }
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
215 # consistency test
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
216 $self->throw("Length of sequence [$seqname] is not [$residuecount]! ")
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
217 unless CORE::length($hash{$count}) == $residuecount;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
218
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
219 $seq = new Bio::LocatableSeq('-seq'=>$hash{$count},
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
220 '-id'=>$seqname,
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
221 '-start'=>$start,
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
222 '-end'=>$end,
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
223 );
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
224
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
225 $aln->add_seq($seq);
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
226
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
227 }
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
228 return $aln;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
229 }
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
230
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
231
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
232 =head2 write_aln
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
233
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
234 Title : write_aln
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
235 Usage : $stream->write_aln(@aln)
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
236 Function: writes the $aln object into the stream in MSF format
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
237 Returns : 1 for success and 0 for error
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
238 Args : L<Bio::Align::AlignI> object
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
239
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
240 =cut
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
241
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
242 sub write_aln {
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
243 my ($self,@aln) = @_;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
244 my $count = 0;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
245 my $wrapped = 0;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
246 my $maxname;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
247 my ($length,$date,$name,$seq,$miss,$pad,
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
248 %hash,@arr,$tempcount,$index,$idlength);
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
249
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
250 foreach my $aln (@aln) {
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
251 if( ! $aln || ! $aln->isa('Bio::Align::AlignI') ) {
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
252 $self->warn("Must provide a Bio::Align::AlignI object when calling write_aln");
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
253 next;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
254 }
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
255 $self->throw("All sequences in the alignment must be the same length")
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
256 unless $aln->is_flush(1) ;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
257
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
258 $aln->set_displayname_flat(); # plain
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
259 $length = $aln->length();
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
260 $self->_print (sprintf(" %s %s\n", $aln->no_sequences, $aln->length));
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
261
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
262 $idlength = $self->idlength();
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
263 foreach $seq ( $aln->each_seq() ) {
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
264 $name = $aln->displayname($seq->get_nse);
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
265 $name = substr($name, 0, $idlength) if length($name) > $idlength;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
266 $name = sprintf("%-".$idlength."s",$name);
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
267 if( $self->interleaved() ) {
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
268 $name .= ' ' ;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
269 } elsif( $self->id_linebreak) {
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
270 $name .= "\n";
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
271 }
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
272
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
273 #phylip needs dashes not dots
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
274 my $seq = $seq->seq();
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
275 $seq=~s/\./-/g;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
276 $hash{$name} = $seq;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
277 push(@arr,$name);
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
278 }
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
279
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
280 if( $self->interleaved() ) {
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
281 while( $count < $length ) {
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
282
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
283 # there is another block to go!
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
284 foreach $name ( @arr ) {
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
285 my $dispname = $name;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
286 $dispname = '' if $wrapped;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
287 $self->_print (sprintf("%".($idlength+3)."s",$dispname));
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
288 $tempcount = $count;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
289 $index = 0;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
290 while( ($tempcount + $idlength < $length) && ($index < 5) ) {
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
291 $self->_print (sprintf("%s ",substr($hash{$name},
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
292 $tempcount,
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
293 $idlength)));
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
294 $tempcount += $idlength;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
295 $index++;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
296 }
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
297 # last
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
298 if( $index < 5) {
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
299 # space to print!
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
300 $self->_print (sprintf("%s ",substr($hash{$name},
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
301 $tempcount)));
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
302 $tempcount += $idlength;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
303 }
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
304 $self->_print ("\n");
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
305 }
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
306 $self->_print ("\n");
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
307 $count = $tempcount;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
308 $wrapped = 1;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
309 }
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
310 } else {
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
311 foreach $name ( @arr ) {
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
312 my $dispname = $name;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
313 $dispname = '' if $wrapped;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
314 $self->_print (sprintf("%s%s\n",$dispname,$hash{$name}));
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
315 }
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
316 }
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
317 }
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
318 $self->flush if $self->_flush_on_write && defined $self->_fh;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
319 return 1;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
320 }
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
321
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
322 =head2 interleaved
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
323
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
324 Title : interleaved
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
325 Usage : my $interleaved = $obj->interleaved
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
326 Function: Get/Set Interleaved status
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
327 Returns : boolean
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
328 Args : boolean
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
329
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
330
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
331 =cut
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
332
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
333 sub interleaved{
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
334 my ($self,$value) = @_;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
335 my $previous = $self->{'_interleaved'};
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
336 if( defined $value ) {
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
337 $self->{'_interleaved'} = $value;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
338 }
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
339 return $previous;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
340 }
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
341
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
342 =head2 idlength
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
343
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
344 Title : idlength
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
345 Usage : my $idlength = $obj->interleaved
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
346 Function: Get/Set value of id length
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
347 Returns : string
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
348 Args : string
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
349
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
350
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
351 =cut
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
352
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
353 sub idlength {
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
354 my($self,$value) = @_;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
355 if (defined $value){
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
356 $self->{'_idlength'} = $value;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
357 }
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
358 return $self->{'_idlength'};
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
359 }
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
360
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
361 =head2 line_length
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
362
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
363 Title : line_length
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
364 Usage : $obj->line_length($newval)
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
365 Function:
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
366 Returns : value of line_length
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
367 Args : newvalue (optional)
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
368
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
369
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
370 =cut
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
371
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
372 sub line_length{
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
373 my ($self,$value) = @_;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
374 if( defined $value) {
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
375 $self->{'line_length'} = $value;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
376 }
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
377 return $self->{'line_length'} || $DEFAULTLINELEN;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
378
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
379 }
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
380
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
381 =head2 id_linebreak
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
382
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
383 Title : id_linebreak
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
384 Usage : $obj->id_linebreak($newval)
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
385 Function:
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
386 Returns : value of id_linebreak
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
387 Args : newvalue (optional)
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
388
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
389
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
390 =cut
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
391
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
392 sub id_linebreak{
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
393 my ($self,$value) = @_;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
394 if( defined $value) {
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
395 $self->{'_id_linebreak'} = $value;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
396 }
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
397 return $self->{'_id_linebreak'} || 0;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
398 }
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
399
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
400 1;