annotate variant_effect_predictor/Bio/EnsEMBL/IndividualSliceFactory.pm @ 3:d30fa12e4cc5 default tip

Merge heads 2:a5976b2dce6f and 1:09613ce8151e which were created as a result of a recently fixed bug.
author devteam <devteam@galaxyproject.org>
date Mon, 13 Jan 2014 10:38:30 -0500
parents 1f6dce3d34e0
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
1 =head1 LICENSE
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
2
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
3 Copyright (c) 1999-2012 The European Bioinformatics Institute and
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
4 Genome Research Limited. All rights reserved.
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
5
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
6 This software is distributed under a modified Apache license.
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
7 For license details, please see
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
8
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
9 http://www.ensembl.org/info/about/code_licence.html
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
10
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
11 =head1 CONTACT
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
12
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
13 Please email comments or questions to the public Ensembl
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
14 developers list at <dev@ensembl.org>.
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
15
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
16 Questions may also be sent to the Ensembl help desk at
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
17 <helpdesk@ensembl.org>.
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
18
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
19 =cut
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
20
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
21 package Bio::EnsEMBL::IndividualSliceFactory;
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
22
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
23 use strict;
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
24 use Bio::EnsEMBL::Utils::Argument qw(rearrange);
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
25 use Bio::EnsEMBL::Utils::Sequence qw(reverse_comp);
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
26 use Bio::EnsEMBL::Slice;
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
27 use Bio::EnsEMBL::Mapper;
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
28 use Bio::EnsEMBL::Utils::Exception qw(throw deprecate warning);
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
29 use Scalar::Util qw(weaken);
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
30
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
31 =head2 new
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
32 =cut
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
33
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
34 sub new{
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
35 my $caller = shift;
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
36 my $class = ref($caller) || $caller;
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
37
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
38 #creates many IndividualSlice objects from the Population
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
39
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
40 my ($population_name, $coord_system, $start, $end, $strand, $seq_region_name, $seq_region_length, $adaptor) = rearrange(['POPULATION', 'COORD_SYSTEM','START','END','STRAND','SEQ_REGION_NAME','SEQ_REGION_LENGTH', 'ADAPTOR'],@_);
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
41
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
42 my $self = bless {
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
43 population_name => $population_name,
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
44 coord_system => $coord_system,
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
45 start => $start,
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
46 end => $end,
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
47 strand => $strand,
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
48 seq_region_name => $seq_region_name,
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
49 seq_region_length => $seq_region_length},$class;
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
50
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
51 $self->adaptor($adaptor);
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
52 return $self;
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
53 }
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
54
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
55 sub adaptor {
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
56 my $self = shift;
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
57
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
58 if(@_) {
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
59 my $ad = shift;
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
60 if($ad && (!ref($ad) || !$ad->isa('Bio::EnsEMBL::DBSQL::BaseAdaptor'))) {
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
61 throw('Adaptor argument must be a Bio::EnsEMBL::DBSQL::BaseAdaptor');
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
62 }
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
63 weaken($self->{'adaptor'} = $ad);
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
64 }
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
65
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
66 return $self->{'adaptor'}
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
67 }
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
68
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
69 sub get_all_IndividualSlice{
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
70 my $self = shift;
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
71
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
72 my $slice;
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
73 if(!$self->adaptor) {
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
74 warning('Cannot get IndividualSlice features without attached adaptor');
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
75 return '';
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
76 }
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
77 my $variation_db = $self->adaptor->db->get_db_adaptor('variation');
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
78
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
79 unless($variation_db) {
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
80 warning("Variation database must be attached to core database to " .
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
81 "retrieve variation information" );
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
82 return '';
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
83 }
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
84 #get the AlleleFeatures in the Population
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
85 my $af_adaptor = $variation_db->get_AlleleFeatureAdaptor;
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
86
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
87 if( $af_adaptor ) {
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
88 #set the adaptor to retrieve data from genotype table
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
89 $af_adaptor->from_IndividualSlice(1);
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
90 #get the Individual for the given strain
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
91 my $population_adaptor = $variation_db->get_PopulationAdaptor;
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
92 my $individual_adaptor = $variation_db->get_IndividualAdaptor;
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
93 if ($population_adaptor && $individual_adaptor){
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
94 $slice = Bio::EnsEMBL::Slice->new(-coord_system => $self->{'coord_system'},
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
95 -start => $self->{'start'},
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
96 -end => $self->{'end'},
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
97 -strand => $self->{'strand'},
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
98 -seq_region_name => $self->{'seq_region_name'},
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
99 -seq_region_length => $self->{'seq_region_length'},
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
100 -adaptor => $self->adaptor
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
101 );
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
102 my $population = $population_adaptor->fetch_by_name($self->{'population_name'});
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
103 #check that there is such population in the database
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
104 if (defined $population){
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
105 #get all the AlleleFeatures in the $population and the Slice given
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
106 my $allele_features = $af_adaptor->fetch_all_by_Slice($slice,$population);
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
107 #get Individuals in the Population
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
108 my $individuals = $individual_adaptor->fetch_all_by_Population($population);
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
109 return $self->_rearrange_Individuals_Alleles($individuals,$allele_features);
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
110 }
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
111 else{
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
112 warning("Population not in the database");
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
113 return '';
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
114
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
115 }
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
116 }
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
117 else{
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
118 warning("Not possible to retrieve PopulationAdaptor from the variation database");
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
119 return '';
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
120 }
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
121 }
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
122
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
123 else{
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
124 warning("Not possible to retrieve AlleleFeatureAdaptor from variation database");
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
125 return '';
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
126 }
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
127 }
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
128
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
129 sub _rearrange_Individuals_Alleles{
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
130 my $self = shift;
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
131 my $individuals = shift;
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
132 my $allele_features;
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
133 my $individual_slice;
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
134 #create the hash with all the individuals
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
135 my %individuals_ids;
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
136 #foreach of the individual, create the IndividualSlice object and add it to the mapping hash
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
137 foreach my $individual (@{$individuals}){
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
138 $individual_slice = Bio::EnsEMBL::Variation::IndividualSlice->new(
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
139 -coord_system => $self->{'coord_system'},
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
140 -start => $self->{'$start'},
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
141 -end => $self->{'end'},
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
142 -strand => $self->{'strand'},
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
143 -seq_region_name => $self->{'seq_region_name'},
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
144 -seq_region_length => $self->{'seq_region_length'},
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
145 -individual => $individual->name);
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
146
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
147 $individuals_ids{$individual->dbID} = $individual_slice;
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
148 }
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
149
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
150 #and rearrange all the AlleleFeatures to the individuals
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
151 foreach my $allele_feature (@{$allele_features}){
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
152 $individuals_ids{$allele_feature->{'_sample_id'}}->add_AlleleFeature($allele_feature);
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
153 }
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
154 my @result = values %individuals_ids;
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
155 return \@result;
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
156 }
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
157
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
158
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
159 1;