-
Notifications
You must be signed in to change notification settings - Fork 117
/
PostGAP.pm
249 lines (179 loc) · 7.34 KB
/
PostGAP.pm
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
=head1 LICENSE
Copyright [1999-2015] Wellcome Trust Sanger Institute and the EMBL-European Bioinformatics Institute
Copyright [2016-2024] EMBL-European Bioinformatics Institute
Licensed under the Apache License, Version 2.0 (the "License");
you may not use this file except in compliance with the License.
You may obtain a copy of the License at
https://www.apache.org/licenses/LICENSE-2.0
Unless required by applicable law or agreed to in writing, software
distributed under the License is distributed on an "AS IS" BASIS,
WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
See the License for the specific language governing permissions and
limitations under the License.
=head1 CONTACT
Ensembl <https://www.ensembl.org/info/about/contact/index.html>
=cut
=head1 NAME
PostGAP - Ensembl Post-GWAS analysis pipeline (PostGAP) - Add PostGAP data fields to the VEP output
=head1 SYNOPSIS
mv PostGAP.pm ~/.vep/Plugins
./vep -i variations.vcf --plugin PostGAP,/path/to/PostGap.gz,col1,col2
=head1 DESCRIPTION
A VEP plugin that retrieves data for variants from a tabix-indexed PostGAP file (1-based file).
Please refer to the PostGAP github and wiki for more information:
https://github.com/Ensembl/postgap
https://github.com/Ensembl/postgap/wiki
https://github.com/Ensembl/postgap/wiki/algorithm-pseudo-code
The Bio::DB::HTS perl library or tabix utility must be installed in your path
to use this plugin. The PostGAP data file can be downloaded from
https://storage.googleapis.com/postgap-data.
The file must be processed and indexed by tabix before use by this plugin.
PostGAP has coordinates for both GRCh38 and GRCh37; the file must be
processed differently according to the assembly you use.
> wget https://storage.googleapis.com/postgap-data/postgap.txt.gz
> gunzip postgap.txt.gz
# GRCh38
> (grep ^"ld_snp_rsID" postgap.txt; grep -v ^"ld_snp_rsID" postgap.txt | sort -k4,4 -k5,5n ) | bgzip > postgap_GRCh38.txt.gz
> tabix -s 4 -b 5 -e 5 -c l postgap_GRCh38.txt.gz
# GRCh37
> (grep ^"ld_snp_rsID" postgap.txt; grep -v ^"ld_snp_rsID" postgap.txt | sort -k2,2 -k3,3n ) | bgzip > postgap_GRCh37.txt.gz
> tabix -s 2 -b 3 -e 3 -c l postgap_GRCh37.txt.gz
Note that in the last command we tell tabix that the header line starts with "l";
this may change to the default of "#" in future versions of PostGAP.
When running the plugin by default 'disease_efo_id', 'disease_name', 'gene_id'
and 'score' information is returned e.g.
--plugin POSTGAP,/path/to/PostGap.gz
You may include all columns with ALL; this fetches a large amount of data per
variant!:
--plugin POSTGAP,/path/to/PostGap.gz,ALL
You may want to select only a specific subset of additional information to be
reported, you can do this by specifying the columns as parameters to the plugin e.g.
--plugin POSTGAP,/path/to/PostGap.gz,gwas_pmid,gwas_size
If a requested column is not found, the error message will report the
complete list of available columns in the POSTGAP file. For a brief description
of the available information please refer to the 'How do I use POSTGAP output?'
section in the POSTGAP wiki.
Tabix also allows the data file to be hosted on a remote server. This plugin is
fully compatible with such a setup - simply use the URL of the remote file:
--plugin PostGAP,http://my.files.com/postgap.txt.gz
Note that gene sequences referred to in PostGAP may be out of sync with
those in the latest release of Ensembl; this may lead to discrepancies with
scores retrieved from other sources.
=cut
package PostGAP;
use strict;
use warnings;
use Bio::EnsEMBL::Variation::Utils::BaseVepTabixPlugin;
use base qw(Bio::EnsEMBL::Variation::Utils::BaseVepTabixPlugin);
my @fields_order;
my $out_txt = 1;
my $out_vcf = 0;
my $out_json = 0;
my $char_sep = "|";
sub new {
my $class = shift;
my $self = $class->SUPER::new(@_);
$self->expand_left(0);
$self->expand_right(0);
my $params_hash = $self->params_to_hash();
# get PostGAP file
my $file = $self->params->[0];
$self->add_file($file);
# get output format
$out_vcf = 1 if ($self->{config}->{output_format} eq "vcf");
$out_json = 1 if ($self->{config}->{output_format} eq "json");
$out_txt = 0 if ($out_vcf || $out_json);
$char_sep = "+" if $out_vcf;
# get headers
open HEAD, "tabix -fh $file 1:1-1 2>&1 | ";
while(<HEAD>) {
chomp;
$self->{headers} = [split];
}
close HEAD;
die "ERROR: Could not read headers from $file\n" unless defined($self->{headers}) && scalar @{$self->{headers}};
my $nDefaultColumns = 4;
# get required columns
my $all =0;
my $i = 1;
while(defined($self->params->[$i])) {
my $col = $self->params->[$i];
if($col eq 'ALL') {
$i = $nDefaultColumns + 1;
$self->{cols} = {map {$_ => $i++}
grep {!defined($self->{cols}->{$_})} #only the extra columns
@{$self->{headers}}};
last; #if ALL is used, then the loop will exit after all existing header elements have been selected
}
die "ERROR: Column $col not found in header for file $file. Available columns are:\n".join(",", @{$self->{headers}})."\n" unless grep {$_ eq $col} @{$self->{headers}};
$self->{cols}->{$self->params->[$i]} = $nDefaultColumns + $i;
$i++;
}
$i += $nDefaultColumns; #ensure that $i is higher than the number of selected columns
#default columns always reported
$self->{cols}->{'disease_efo_id'} = 1;
$self->{cols}->{'disease_name'} = 2;
$self->{cols}->{'gene_id'} = 3;
$self->{cols}->{'score'} = 4;
# get the order of the output fields into an array, $i is the total number of columns +1
@fields_order = map { $_ }
sort {
(defined($self->{cols}->{$a}) ? $self->{cols}->{$a} : $i)
<=>
(defined($self->{cols}->{$b}) ? $self->{cols}->{$b} : $i)
||
$a cmp $b
}
keys %{$self->{cols}};
return $self;
}
sub feature_types {
return ['Feature','Intergenic'];
}
sub get_header_info {
my $self = shift;
my $header = 'PostGAP data for variation - phenotype association. Format: ';
$header .= join($char_sep, @fields_order );
return {
PostGAP => $header,
}
}
sub run {
my $self = shift;
my $tva = shift;
my $vf = $tva->variation_feature;
my ($start, $end) = ($vf->{start}, $vf->{end});
# adjust coords for insertions
($start, $end) = ($vf->{end}, $vf->{start}) if ($vf->{start} > $vf->{end});
my $data = $self->get_data($vf->{chr}, $start, $end);
return {} unless $data && scalar @$data;
my @result =();
my %result_uniq;
my @result_str = ();
foreach my $tmp_data(@{$data}) {
# get required data
my %tmp_return =
map {$_ => $tmp_data->{$_}}
grep {defined($self->{cols}->{$_})} # only include selected cols
keys %$tmp_data;
# get only 3 significant digits
$tmp_return{score} = sprintf("%.3f", $tmp_return{score}) if defined $tmp_return{score};
# report only unique set of fields
my $record_line = join(",", values %tmp_return);
next if defined $result_uniq{$record_line};
$result_uniq{$record_line} = 1;
push(@result_str, join($char_sep, @tmp_return{@fields_order}));
push(@result, \%tmp_return);
}
return {
PostGAP => $self->{config}->{output_format} eq "json" ? \@result : \@result_str
}
}
sub parse_data {
my ($self, $line) = @_;
my @split = split /\t/, $line;
# parse data into hash of col names and values
my %data = map {$self->{headers}->[$_] => $split[$_]} (0..(scalar @{$self->{headers}} - 1));
return \%data;
}
1;