annotate variant_effect_predictor/Bio/Align/PairwiseStatistics.pm @ 0:21066c0abaf5 draft

Uploaded
author willmclaren
date Fri, 03 Aug 2012 10:04:48 -0400
parents
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
1 # $Id: PairwiseStatistics.pm,v 1.2 2002/10/22 07:45:10 lapp Exp $
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
2 #
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
3 # BioPerl module for Bio::Align::PairwiseStatistics
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
4 #
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
5 # Cared for by Jason Stajich <jason@bioperl.org>
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
6 #
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
7 # Copyright Jason Stajich
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
8 #
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
9 # You may distribute this module under the same terms as perl itself
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
10
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
11 # POD documentation - main docs before the code
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
12
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
13 =head1 NAME
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
14
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
15 Bio::Align::PairwiseStatistics - Base statistic object for Pairwise Alignments
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
16
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
17 =head1 SYNOPSIS
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
18
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
19 Give standard usage here
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
20
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
21 =head1 DESCRIPTION
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
22
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
23 Describe the object here
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
24
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
25 =head1 FEEDBACK
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
26
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
27 =head2 Mailing Lists
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
28
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
29 User feedback is an integral part of the evolution of this and other
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
30 Bioperl modules. Send your comments and suggestions preferably to
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
31 the Bioperl mailing list. Your participation is much appreciated.
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
32
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
33 bioperl-l@bioperl.org - General discussion
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
34 http://bioperl.org/MailList.shtml - About the mailing lists
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
35
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
36 =head2 Reporting Bugs
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
37
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
38 Report bugs to the Bioperl bug tracking system to help us keep track
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
39 of the bugs and their resolution. Bug reports can be submitted via
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
40 email or the web:
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
41
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
42 bioperl-bugs@bioperl.org
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
43 http://bugzilla.bioperl.org/
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
44
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
45 =head1 AUTHOR - Jason Stajich
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
46
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
47 Email jason@bioperl.org
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
48
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
49 Describe contact details here
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
50
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
51 =head1 CONTRIBUTORS
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
52
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
53 Additional contributors names and emails here
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
54
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
55 =head1 APPENDIX
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
56
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
57 The rest of the documentation details each of the object methods.
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
58 Internal methods are usually preceded with a _
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
59
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
60 =cut
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
61
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
62
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
63 # Let the code begin...
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
64
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
65
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
66 package Bio::Align::PairwiseStatistics;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
67 use vars qw(@ISA $GapChars);
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
68 use strict;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
69
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
70 use Bio::Align::StatisticsI;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
71 use Bio::Root::Root;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
72
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
73 BEGIN { $GapChars = '(\.|\-)'; }
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
74
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
75 @ISA = qw(Bio::Root::Root Bio::Align::StatisticsI );
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
76
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
77 =head2 number_of_comparable_bases
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
78
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
79 Title : number_of_comparable_bases
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
80 Usage : my $bases = $stat->number_of_comparable_bases($aln);
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
81 Function: Returns the count of the number of bases that can be
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
82 compared (L) in this alignment ( length - gaps)
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
83 Returns : integer
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
84 Args : Bio::Align::AlignI
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
85
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
86
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
87 =cut
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
88
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
89 sub number_of_comparable_bases{
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
90 my ($self,$aln) = @_;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
91 if( ! defined $aln || ! $aln->isa('Bio::Align::AlignI') ) {
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
92 $self->warn("Must provide a Bio::Align::AlignI compliant object to Bio::Align::PairwiseStatistics");
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
93 return 0;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
94 } elsif( $aln->no_sequences != 2 ) {
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
95 $self->warn("only pairwise calculations currently supported");
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
96 }
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
97 my $L = $aln->length - $self->number_of_gaps($aln);
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
98 return $L;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
99 }
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
100
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
101 =head2 number_of_differences
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
102
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
103 Title : number_of_differences
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
104 Usage : my $nd = $stat->number_of_distances($aln);
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
105 Function: Returns the number of differences between two
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
106 Returns : integer
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
107 Args : Bio::Align::AlignI
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
108
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
109
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
110 =cut
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
111
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
112 sub number_of_differences{
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
113 my ($self,$aln) = @_;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
114 if( ! defined $aln || ! $aln->isa('Bio::Align::AlignI') ) {
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
115 $self->warn("Must provide a Bio::Align::AlignI compliant object to Bio::Align::PairwiseStatistics");
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
116 return 0;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
117 } elsif( $aln->no_sequences != 2 ) {
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
118 $self->warn("only pairwise calculations currently supported");
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
119 }
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
120 my (@seqs);
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
121 foreach my $seq ( $aln->each_seq) {
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
122 push @seqs, [ split(//,$seq->seq())];
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
123 }
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
124 my $firstseq = shift @seqs;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
125 # my $secondseq = shift @seqs;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
126 my $diffcount = 0;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
127 for (my $i = 0;$i<$aln->length; $i++ ) {
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
128 next if( $firstseq->[$i] =~ /^$GapChars$/);
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
129 foreach my $seq ( @seqs ) {
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
130 next if( $seq->[$i] =~ /^$GapChars$/);
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
131 if( $firstseq->[$i] ne $seq->[$i] ) {
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
132 $diffcount++;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
133 }
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
134 }
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
135 }
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
136 return $diffcount;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
137 }
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
138
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
139 =head2 number_of_gaps
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
140
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
141 Title : number_of_gaps
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
142 Usage : my $nd = $stat->number_of_gaps($aln);
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
143 Function: Returns the number of differences between two
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
144 Example :
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
145 Returns :
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
146 Args :
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
147
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
148
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
149 =cut
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
150
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
151 sub number_of_gaps{
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
152 my ($self,$aln) = @_;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
153 if( ! defined $aln || ! $aln->isa('Bio::Align::AlignI') ) {
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
154 $self->warn("Must provide a Bio::Align::AlignI compliant object to Bio::Align::PairwiseStatistics");
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
155 return 0;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
156 } elsif( $aln->no_sequences != 2 ) {
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
157 $self->warn("only pairwise calculations currently supported");
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
158 }
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
159 my (@seqs);
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
160 foreach my $seq ( $aln->each_seq) {
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
161 push @seqs, [ split(//,$seq->seq())];
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
162 }
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
163 my $firstseq = shift @seqs;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
164 # my $secondseq = shift @seqs;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
165 my $gapcount = 0;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
166 for (my $i = 0;$i<$aln->length; $i++ ) {
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
167 ($gapcount++) && next if( $firstseq->[$i] =~ /^$GapChars$/);
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
168 foreach my $seq ( @seqs ) {
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
169 ($gapcount++) && next if( $seq->[$i] =~ /^$GapChars$/);
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
170 }
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
171 }
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
172 return $gapcount;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
173 }
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
174
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
175 1;