annotate variant_effect_predictor/Bio/EnsEMBL/Utils/Slice.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 =head1 LICENSE
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
2
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
3 Copyright (c) 1999-2012 The European Bioinformatics Institute and
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
4 Genome Research Limited. All rights reserved.
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
5
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
6 This software is distributed under a modified Apache license.
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
7 For license details, please see
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
8
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
9 http://www.ensembl.org/info/about/code_licence.html
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
10
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
11 =head1 CONTACT
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
12
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
13 Please email comments or questions to the public Ensembl
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
14 developers list at <dev@ensembl.org>.
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
15
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
16 Questions may also be sent to the Ensembl help desk at
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
17 <helpdesk@ensembl.org>.
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
18
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
19 =cut
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
20
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
21 =head1 NAME
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
22
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
23 Bio::EnsEMBL::Utils::Slice - Utility functions for slices
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
24
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
25 =head1 SYNOPSIS
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
26
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
27 use Bio::EnsEMBL::Utils::Slice qw(split_Slices);
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
28
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
29 # ...
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
30
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
31 # get all chromosomes in the database
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
32 my $slices = $slice_adaptor->fetch_all('chromosome');
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
33
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
34 # split the chromosomes into equal chunks of size less than 1MB
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
35 # with an overlap of 1kb
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
36 $slices = split_Slices( $slices, 1e6, 1e3 );
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
37
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
38 =head1 METHODS
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
39
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
40 =cut
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
41
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
42
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
43 package Bio::EnsEMBL::Utils::Slice;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
44
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
45 use strict;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
46 use warnings;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
47
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
48 use Exporter;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
49
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
50 use vars qw(@ISA @EXPORT_OK);
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
51
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
52 @ISA = qw(Exporter);
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
53
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
54 @EXPORT_OK = qw(&split_Slices);
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
55
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
56 use Bio::EnsEMBL::Utils::Exception qw(throw);
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
57 use POSIX;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
58
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
59 =head2 split_Slices
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
60
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
61 Arg [1] : ref to list of slices
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
62 Arg [2] : int maxlength of sub slices
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
63 Arg [3] : int overlap length (optional)
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
64 Example : my $sub_slices = split_Slices($slices,$maxlen,$overlap)
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
65 Description: splits a slice into smaller slices
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
66 Returntype : ref to list of slices
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
67 Exceptions : maxlen <1 or overlap < 0
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
68
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
69 =cut
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
70
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
71 sub split_Slices{
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
72 my ($slice_big,$max_length,$overlap)=@_;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
73
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
74 if(!defined($max_length) or $max_length < 1){
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
75 throw("maxlength needs to be set and > 0");
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
76 }
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
77
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
78 if(!defined($overlap)){
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
79 $overlap = 0;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
80 }
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
81 elsif($overlap < 0){
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
82 throw("negative overlaps not allowed");
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
83 }
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
84
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
85 my @out=();
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
86
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
87 foreach my $slice (@$slice_big){
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
88
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
89 my $start = $slice->start;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
90 my $end;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
91 my $multiple;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
92 my $number;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
93 my $length = $slice->length;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
94
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
95 if($max_length && ($length > $overlap)) {
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
96 #No seq region may be longer than max_length but we want to make
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
97 #them all similar size so that the last one isn't much shorter.
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
98 #Divide the seq_region into the largest equal pieces that are shorter
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
99 #than max_length
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
100
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
101 #calculate number of slices to create
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
102 $number = ($length-$overlap) / ($max_length-$overlap);
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
103 $number = ceil($number); #round up to int
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
104
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
105 #calculate length of created slices
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
106 $multiple = $length / $number;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
107 $multiple = floor($multiple); #round down to int
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
108 } else {
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
109 #just one slice of the whole seq_region
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
110 $number = 1;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
111 $multiple = $length;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
112 }
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
113
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
114 my $i;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
115 for(my $i=0; $i < $number; $i++) {
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
116 $end = $start + $multiple + $overlap;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
117
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
118 #any remainder gets added to the last slice of the seq_region
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
119 $end = $slice->end if($i == $number-1);
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
120 push @out, Bio::EnsEMBL::Slice->new
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
121 (-START => $start,
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
122 -END => $end,
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
123 -STRAND => 1,
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
124 -SEQ_REGION_NAME => $slice->seq_region_name,
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
125 -SEQ_REGION_LENGTH => $slice->seq_region_length,
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
126 -COORD_SYSTEM => $slice->coord_system,
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
127 -ADAPTOR => $slice->adaptor);
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
128 $start += $multiple + 1;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
129 }
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
130 }
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
131
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
132 return \@out;
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
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
137 1;