|
0
|
1 # $Id: msf.pm,v 1.16 2002/11/26 16:34:39 jason Exp $
|
|
|
2 #
|
|
|
3 # BioPerl module for Bio::AlignIO::msf
|
|
|
4
|
|
|
5 # based on the Bio::SeqIO::msf module
|
|
|
6 # by Ewan Birney <birney@sanger.ac.uk>
|
|
|
7 # and Lincoln Stein <lstein@cshl.org>
|
|
|
8 #
|
|
|
9 # and the SimpleAlign.pm module of Ewan Birney
|
|
|
10 #
|
|
|
11 # Copyright Peter Schattner
|
|
|
12 #
|
|
|
13 # You may distribute this module under the same terms as perl itself
|
|
|
14 # _history
|
|
|
15 # September 5, 2000
|
|
|
16 # POD documentation - main docs before the code
|
|
|
17
|
|
|
18 =head1 NAME
|
|
|
19
|
|
|
20 Bio::AlignIO::msf - msf sequence input/output stream
|
|
|
21
|
|
|
22 =head1 SYNOPSIS
|
|
|
23
|
|
|
24 Do not use this module directly. Use it via the L<Bio::AlignIO> class.
|
|
|
25
|
|
|
26 =head1 DESCRIPTION
|
|
|
27
|
|
|
28 This object can transform L<Bio::Align::AlignI> objects to and from msf flat
|
|
|
29 file databases.
|
|
|
30
|
|
|
31 =head1 FEEDBACK
|
|
|
32
|
|
|
33 =head2 Reporting Bugs
|
|
|
34
|
|
|
35 Report bugs to the Bioperl bug tracking system to help us keep track
|
|
|
36 the bugs and their resolution.
|
|
|
37 Bug reports can be submitted via email or the web:
|
|
|
38
|
|
|
39 bioperl-bugs@bio.perl.org
|
|
|
40 http://bugzilla.bioperl.org/
|
|
|
41
|
|
|
42 =head1 AUTHORS - Peter Schattner
|
|
|
43
|
|
|
44 Email: schattner@alum.mit.edu
|
|
|
45
|
|
|
46
|
|
|
47 =head1 APPENDIX
|
|
|
48
|
|
|
49 The rest of the documentation details each of the object
|
|
|
50 methods. Internal methods are usually preceded with a _
|
|
|
51
|
|
|
52 =cut
|
|
|
53
|
|
|
54 # Let the code begin...
|
|
|
55
|
|
|
56 package Bio::AlignIO::msf;
|
|
|
57 use vars qw(@ISA %valid_type);
|
|
|
58 use strict;
|
|
|
59
|
|
|
60 use Bio::AlignIO;
|
|
|
61 use Bio::SeqIO::gcg; # for GCG_checksum()
|
|
|
62 use Bio::SimpleAlign;
|
|
|
63
|
|
|
64 @ISA = qw(Bio::AlignIO);
|
|
|
65
|
|
|
66 BEGIN {
|
|
|
67 %valid_type = qw( dna N rna N protein P );
|
|
|
68 }
|
|
|
69
|
|
|
70 =head2 next_aln
|
|
|
71
|
|
|
72 Title : next_aln
|
|
|
73 Usage : $aln = $stream->next_aln()
|
|
|
74 Function: returns the next alignment in the stream. Tries to read *all* MSF
|
|
|
75 It reads all non whitespace characters in the alignment
|
|
|
76 area. For MSFs with weird gaps (eg ~~~) map them by using
|
|
|
77 $al->map_chars('~','-')
|
|
|
78 Returns : L<Bio::Align::AlignI> object
|
|
|
79 Args : NONE
|
|
|
80
|
|
|
81 =cut
|
|
|
82
|
|
|
83 sub next_aln {
|
|
|
84 my $self = shift;
|
|
|
85 my $entry;
|
|
|
86 my (%hash,$name,$str,@names,$seqname,$start,$end,$count,$seq);
|
|
|
87
|
|
|
88 my $aln = Bio::SimpleAlign->new(-source => 'gcg' );
|
|
|
89
|
|
|
90 while( $entry = $self->_readline) {
|
|
|
91 $entry =~ /\/\// && last; # move to alignment section
|
|
|
92 $entry =~ /Name:\s+(\S+)/ && do { $name = $1;
|
|
|
93 $hash{$name} = ""; # blank line
|
|
|
94 push(@names,$name); # we need it ordered!
|
|
|
95 };
|
|
|
96 # otherwise - skip
|
|
|
97 }
|
|
|
98
|
|
|
99 # alignment section
|
|
|
100
|
|
|
101 while( $entry = $self->_readline) {
|
|
|
102 next if ( $entry =~ /^\s+(\d+)/ ) ;
|
|
|
103 $entry =~ /^\s*(\S+)\s+(.*)$/ && do {
|
|
|
104 $name = $1;
|
|
|
105 $str = $2;
|
|
|
106 if( ! exists $hash{$name} ) {
|
|
|
107 $self->throw("$name exists as an alignment line but not in the header. Not confident of what is going on!");
|
|
|
108 }
|
|
|
109 $str =~ s/\s//g;
|
|
|
110 $hash{$name} .= $str;
|
|
|
111 };
|
|
|
112 }
|
|
|
113
|
|
|
114 return 0 if scalar @names < 1;
|
|
|
115
|
|
|
116 # now got this as a name - sequence hash. Lets make some sequences!
|
|
|
117
|
|
|
118 foreach $name ( @names ) {
|
|
|
119 if( $name =~ /(\S+)\/(\d+)-(\d+)/ ) {
|
|
|
120 $seqname = $1;
|
|
|
121 $start = $2;
|
|
|
122 $end = $3;
|
|
|
123 } else {
|
|
|
124 $seqname=$name;
|
|
|
125 $start = 1;
|
|
|
126 $str = $hash{$name};
|
|
|
127 $str =~ s/[^A-Za-z]//g;
|
|
|
128 $end = length($str);
|
|
|
129 }
|
|
|
130
|
|
|
131 $seq = new Bio::LocatableSeq('-seq'=>$hash{$name},
|
|
|
132 '-id'=>$seqname,
|
|
|
133 '-start'=>$start,
|
|
|
134 '-end'=>$end,
|
|
|
135 );
|
|
|
136
|
|
|
137 $aln->add_seq($seq);
|
|
|
138
|
|
|
139
|
|
|
140 # If $end <= 0, we have either reached the end of
|
|
|
141 # file in <> or we have encountered some other error
|
|
|
142 #
|
|
|
143 # if ($end <= 0) { undef $aln;}
|
|
|
144
|
|
|
145
|
|
|
146 }
|
|
|
147
|
|
|
148 return $aln;
|
|
|
149 }
|
|
|
150
|
|
|
151
|
|
|
152
|
|
|
153
|
|
|
154 =head2 write_aln
|
|
|
155
|
|
|
156 Title : write_aln
|
|
|
157 Usage : $stream->write_aln(@aln)
|
|
|
158 Function: writes the $aln object into the stream in MSF format
|
|
|
159 Sequence type of the alignment is determined by the first sequence.
|
|
|
160 Returns : 1 for success and 0 for error
|
|
|
161 Args : L<Bio::Align::AlignI> object
|
|
|
162
|
|
|
163
|
|
|
164 =cut
|
|
|
165
|
|
|
166 sub write_aln {
|
|
|
167 my ($self,@aln) = @_;
|
|
|
168 my $msftag;
|
|
|
169 my $type;
|
|
|
170 my $count = 0;
|
|
|
171 my $maxname;
|
|
|
172 my ($length,$date,$name,$seq,$miss,$pad,%hash,@arr,$tempcount,$index);
|
|
|
173 foreach my $aln (@aln) {
|
|
|
174 if( ! $aln || ! $aln->isa('Bio::Align::AlignI') ) {
|
|
|
175 $self->warn("Must provide a Bio::Align::AlignI object when calling write_aln");
|
|
|
176 next;
|
|
|
177 }
|
|
|
178 $date = localtime(time);
|
|
|
179 $msftag = "MSF";
|
|
|
180 $type = $valid_type{$aln->get_seq_by_pos(1)->alphabet};
|
|
|
181 $maxname = $aln->maxdisplayname_length();
|
|
|
182 $length = $aln->length();
|
|
|
183 $name = $aln->id();
|
|
|
184 if( !defined $name ) {
|
|
|
185 $name = "Align";
|
|
|
186 }
|
|
|
187
|
|
|
188
|
|
|
189 $self->_print (sprintf("\n%s MSF: %d Type: %s %s Check: 00 ..\n\n",
|
|
|
190 $name, $aln->no_sequences, $type, $date));
|
|
|
191
|
|
|
192
|
|
|
193 foreach $seq ( $aln->each_seq() ) {
|
|
|
194
|
|
|
195
|
|
|
196 $name = $aln->displayname($seq->get_nse());
|
|
|
197 $miss = $maxname - length ($name);
|
|
|
198 $miss += 2;
|
|
|
199 $pad = " " x $miss;
|
|
|
200
|
|
|
201 $self->_print (sprintf(" Name: %s%sLen: %d Check: %d Weight: 1.00\n",$name,$pad,length $seq->seq(), Bio::SeqIO::gcg->GCG_checksum($seq)));
|
|
|
202
|
|
|
203 $hash{$name} = $seq->seq();
|
|
|
204 push(@arr,$name);
|
|
|
205 }
|
|
|
206 # ok - heavy handed, but there you go.
|
|
|
207 #
|
|
|
208 $self->_print ("\n//\n\n\n");
|
|
|
209
|
|
|
210 while( $count < $length ) {
|
|
|
211 # there is another block to go!
|
|
|
212 foreach $name ( @arr ) {
|
|
|
213 $self->_print (sprintf("%-20s ",$name));
|
|
|
214
|
|
|
215 $tempcount = $count;
|
|
|
216 $index = 0;
|
|
|
217 while( ($tempcount + 10 < $length) && ($index < 5) ) {
|
|
|
218
|
|
|
219 $self->_print (sprintf("%s ",substr($hash{$name},$tempcount,10)));
|
|
|
220
|
|
|
221 $tempcount += 10;
|
|
|
222 $index++;
|
|
|
223 } #
|
|
|
224 # ok, could be the very last guy ;)
|
|
|
225 #
|
|
|
226 if( $index < 5) {
|
|
|
227 # space to print!
|
|
|
228 #
|
|
|
229 $self->_print (sprintf("%s ",substr($hash{$name},$tempcount)));
|
|
|
230 $tempcount += 10;
|
|
|
231 }
|
|
|
232 $self->_print ("\n");
|
|
|
233 }
|
|
|
234 $self->_print ("\n\n");
|
|
|
235 $count = $tempcount;
|
|
|
236 }
|
|
|
237 }
|
|
|
238 $self->flush if $self->_flush_on_write && defined $self->_fh;
|
|
|
239 return 1;
|
|
|
240 }
|
|
|
241
|
|
|
242 1;
|
