annotate dir_plugins/NearestGene.pm @ 10:f594c6bed58f draft default tip

Uploaded
author dvanzessen
date Tue, 21 Apr 2020 11:40:19 +0000
parents e545d0a25ffe
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
1 =head1 LICENSE
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
2
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
3 Copyright [1999-2015] Wellcome Trust Sanger Institute and the EMBL-European Bioinformatics Institute
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
4 Copyright [2016-2018] EMBL-European Bioinformatics Institute
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
5
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
6 Licensed under the Apache License, Version 2.0 (the "License");
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
7 you may not use this file except in compliance with the License.
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
8 You may obtain a copy of the License at
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
9
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
10 http://www.apache.org/licenses/LICENSE-2.0
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
11
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
12 Unless required by applicable law or agreed to in writing, software
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
13 distributed under the License is distributed on an "AS IS" BASIS,
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
14 WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
15 See the License for the specific language governing permissions and
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
16 limitations under the License.
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
17
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
18 =head1 CONTACT
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
19
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
20 Ensembl <http://www.ensembl.org/info/about/contact/index.html>
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
21
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
22 =cut
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
23
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
24 =head1 NAME
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
25
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
26 NearestGene
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
27
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
28 =head1 SYNOPSIS
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
29
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
30 mv NearestGene.pm ~/.vep/Plugins
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
31 ./vep -i variations.vcf --cache --plugin NearestGene
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
32
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
33 =head1 DESCRIPTION
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
34
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
35 This is a plugin for the Ensembl Variant Effect Predictor (VEP) that
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
36 finds the nearest gene(s) to a non-genic variant. More than one gene
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
37 may be reported if the genes overlap the variant or if genes are
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
38 equidistant.
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
39
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
40 Various parameters can be altered by passing them to the plugin command:
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
41
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
42 - limit : limit the number of genes returned (default: 1)
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
43 - range : initial search range in bp (default: 1000)
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
44 - max_range : maximum search range in bp (default: 10000)
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
45
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
46 Parameters are passed e.g.:
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
47
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
48 --plugin NearestGene,limit=3,max_range=50000
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
49
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
50 =cut
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
51
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
52 package NearestGene;
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
53
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
54 use strict;
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
55 use warnings;
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
56
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
57 use Bio::EnsEMBL::Variation::Utils::BaseVepPlugin;
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
58
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
59 use base qw(Bio::EnsEMBL::Variation::Utils::BaseVepPlugin);
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
60
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
61 my %CONFIG = (
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
62 limit => 1,
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
63 range => 1000,
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
64 max_range => 10000,
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
65 );
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
66
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
67 sub new {
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
68 my $class = shift;
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
69
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
70 my $self = $class->SUPER::new(@_);
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
71
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
72 my $params = $self->params;
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
73
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
74 foreach my $param(@$params) {
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
75 my ($key, $val) = split('=', $param);
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
76 die("ERROR: Failed to parse parameter $param\n") unless defined($key) && defined($val);
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
77 $CONFIG{$key} = $val;
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
78 }
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
79
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
80 return $self;
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
81 }
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
82
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
83 sub feature_types {
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
84 return ['Intergenic','MotifFeature','RegulatoryFeature'];
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
85 }
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
86
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
87 sub variant_feature_types {
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
88 return ['BaseVariationFeature'];
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
89 }
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
90
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
91 sub get_header_info {
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
92 return {
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
93 NearestGene => "Ensembl identifier of nearest gene"
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
94 };
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
95 }
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
96
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
97 sub run {
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
98 my ($self, $vfoa) = @_;
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
99
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
100 my $vf = $vfoa->base_variation_feature;
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
101 my $loc_string = sprintf("%s:%i-%i", $vf->{chr} || $vf->seq_region_name, $vf->{start}, $vf->{end});
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
102
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
103 if(!exists($self->{_cache}) || !exists($self->{_cache}->{$loc_string})) {
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
104 $self->{config}->{ga} = $self->{config}->{reg}->get_adaptor($self->{config}->{species}, $self->{config}->{core_type}, 'gene');
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
105 $self->{ga} ||= $self->{config}->{ga};
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
106 die("ERROR: Could not get gene adaptor; this plugin does not work in --offline mode\n") unless $self->{ga};
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
107
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
108 my %opts = map {'-'.$_ => $CONFIG{$_}} keys %CONFIG;
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
109 $opts{-feature} = $vf;
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
110
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
111 my @result = map {$_->[0]->stable_id} @{
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
112 $self->{ga}->fetch_all_by_outward_search(%opts)
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
113 };
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
114
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
115 $self->{_cache}->{$loc_string} = scalar @result ? join(",", @result) : undef;
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
116 }
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
117
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
118 return $self->{_cache}->{$loc_string} ? { NearestGene => $self->{_cache}->{$loc_string} } : {};
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
119 }
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
120
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
121 1;
e545d0a25ffe Uploaded
dvanzessen
parents:
diff changeset
122