line |
stmt |
bran |
cond |
sub |
pod |
time |
code |
1
|
|
|
|
|
|
|
package Bio::VertRes::Config::Pipelines::SnpCalling; |
2
|
|
|
|
|
|
|
|
3
|
|
|
|
|
|
|
# ABSTRACT: The base class for the SNP calling pipeline. |
4
|
|
|
|
|
|
|
|
5
|
|
|
|
|
|
|
|
6
|
1
|
|
|
1
|
|
143261
|
use Moose; |
|
0
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
7
|
|
|
|
|
|
|
use Bio::VertRes::Config::Pipelines::Common; |
8
|
|
|
|
|
|
|
use Bio::VertRes::Config::References; |
9
|
|
|
|
|
|
|
use Bio::VertRes::Config::Pipelines::Roles::MetaDataFilter; |
10
|
|
|
|
|
|
|
use Bio::VertRes::Config::Pipelines::Roles::MultiplePrefix; |
11
|
|
|
|
|
|
|
extends 'Bio::VertRes::Config::Pipelines::Common'; |
12
|
|
|
|
|
|
|
with 'Bio::VertRes::Config::Pipelines::Roles::MetaDataFilter'; |
13
|
|
|
|
|
|
|
with 'Bio::VertRes::Config::Pipelines::Roles::MultiplePrefix'; |
14
|
|
|
|
|
|
|
with 'Bio::VertRes::Config::Pipelines::Roles::FilenameWithReference'; |
15
|
|
|
|
|
|
|
|
16
|
|
|
|
|
|
|
has 'pipeline_short_name' => ( is => 'ro', isa => 'Str', default => 'snps' ); |
17
|
|
|
|
|
|
|
has 'module' => ( is => 'ro', isa => 'Str', default => 'VertRes::Pipelines::SNPs' ); |
18
|
|
|
|
|
|
|
has 'reference' => ( is => 'ro', isa => 'Str', required => 1 ); |
19
|
|
|
|
|
|
|
has 'reference_lookup_file' => ( is => 'ro', isa => 'Str', required => 1 ); |
20
|
|
|
|
|
|
|
has 'toplevel_action' => ( is => 'ro', isa => 'Str', default => '__VRTrack_SNPs__' ); |
21
|
|
|
|
|
|
|
has 'run_after_bam_improvement' => ( is => 'ro', isa => 'Bool', default => 0); |
22
|
|
|
|
|
|
|
|
23
|
|
|
|
|
|
|
has '_max_lanes' => ( is => 'ro', isa => 'Int', default => 300 ); |
24
|
|
|
|
|
|
|
has '_pseudo_genome' => ( is => 'ro', isa => 'Bool', default => 1 ); |
25
|
|
|
|
|
|
|
has '_bam_suffix' => ( is => 'ro', isa => 'Str', default => 'markdup.bam' ); |
26
|
|
|
|
|
|
|
|
27
|
|
|
|
|
|
|
has '_bsub_opts' => ( |
28
|
|
|
|
|
|
|
is => 'ro', |
29
|
|
|
|
|
|
|
isa => 'Str', |
30
|
|
|
|
|
|
|
default => "-q normal -M3500000 -R 'select[type==X86_64 && mem>3500] rusage[mem=3500,thouio=1,tmp=16000]'" |
31
|
|
|
|
|
|
|
); |
32
|
|
|
|
|
|
|
has '_bsub_opts_long' => ( |
33
|
|
|
|
|
|
|
is => 'ro', |
34
|
|
|
|
|
|
|
isa => 'Str', |
35
|
|
|
|
|
|
|
default => "-q normal -M3500000 -R 'select[type==X86_64 && mem>3500] rusage[mem=3500,thouio=1,tmp=16000]'" |
36
|
|
|
|
|
|
|
); |
37
|
|
|
|
|
|
|
has '_bsub_opts_mpileup' => |
38
|
|
|
|
|
|
|
( is => 'ro', isa => 'Str', default => "-q normal -R 'select[type==X86_64] rusage[thouio=1]'" ); |
39
|
|
|
|
|
|
|
has '_split_size_mpileup' => ( is => 'ro', isa => 'Int', default => 300000000 ); |
40
|
|
|
|
|
|
|
has '_tmp_dir' => ( is => 'ro', isa => 'Str', default => '/lustre/scratch108/pathogen/tmp' ); |
41
|
|
|
|
|
|
|
has '_mpileup_cmd' => ( is => 'ro', isa => 'Str', default => 'samtools mpileup -d 1000 -DSug ' ); |
42
|
|
|
|
|
|
|
has '_max_jobs' => ( is => 'ro', isa => 'Int', default => 100 ); |
43
|
|
|
|
|
|
|
has '_fai_chr_regex' => ( is => 'ro', isa => 'Str', default => '[\w\.\#]+' ); |
44
|
|
|
|
|
|
|
has '_fa_ref' => ( is => 'ro', isa => 'Str', lazy => 1, builder => '_build__fa_ref' ); |
45
|
|
|
|
|
|
|
has '_fai_ref' => ( is => 'ro', isa => 'Str', lazy => 1, builder => '_build__fai_ref' ); |
46
|
|
|
|
|
|
|
has '_ignore_snp_called_status' => ( is => 'ro', isa => 'Int', default => 1 ); |
47
|
|
|
|
|
|
|
|
48
|
|
|
|
|
|
|
sub _build__fa_ref { |
49
|
|
|
|
|
|
|
my ($self) = @_; |
50
|
|
|
|
|
|
|
Bio::VertRes::Config::References->new( reference_lookup_file => $self->reference_lookup_file ) |
51
|
|
|
|
|
|
|
->get_reference_location_on_disk( $self->reference ); |
52
|
|
|
|
|
|
|
} |
53
|
|
|
|
|
|
|
|
54
|
|
|
|
|
|
|
sub _build__fai_ref { |
55
|
|
|
|
|
|
|
my ($self) = @_; |
56
|
|
|
|
|
|
|
return join( '.', ( $self->_fa_ref, 'fai' ) ); |
57
|
|
|
|
|
|
|
} |
58
|
|
|
|
|
|
|
|
59
|
|
|
|
|
|
|
override 'to_hash' => sub { |
60
|
|
|
|
|
|
|
my ($self) = @_; |
61
|
|
|
|
|
|
|
my $output_hash = super(); |
62
|
|
|
|
|
|
|
|
63
|
|
|
|
|
|
|
$output_hash->{max_lanes} = $self->_max_lanes; |
64
|
|
|
|
|
|
|
$output_hash->{vrtrack_processed_flags} = { import => 1, qc => 1, stored => 1, mapped => 1}; |
65
|
|
|
|
|
|
|
$output_hash->{vrtrack_processed_flags}{improved} = 1 if($self->run_after_bam_improvement); |
66
|
|
|
|
|
|
|
|
67
|
|
|
|
|
|
|
if ( $self->_pseudo_genome ) { |
68
|
|
|
|
|
|
|
$output_hash->{data}{task} = 'pseudo_genome,mpileup,update_db,cleanup'; |
69
|
|
|
|
|
|
|
} |
70
|
|
|
|
|
|
|
else { |
71
|
|
|
|
|
|
|
$output_hash->{data}{task} = 'mpileup,update_db,cleanup'; |
72
|
|
|
|
|
|
|
} |
73
|
|
|
|
|
|
|
$output_hash->{data}{bam_suffix} = $self->_bam_suffix; |
74
|
|
|
|
|
|
|
$output_hash->{data}{bsub_opts} = $self->_bsub_opts; |
75
|
|
|
|
|
|
|
$output_hash->{data}{bsub_opts_long} = $self->_bsub_opts_long; |
76
|
|
|
|
|
|
|
$output_hash->{data}{bsub_opts_mpileup} = $self->_bsub_opts_mpileup; |
77
|
|
|
|
|
|
|
$output_hash->{data}{split_size_mpileup} = $self->_split_size_mpileup; |
78
|
|
|
|
|
|
|
$output_hash->{data}{tmp_dir} = $self->_tmp_dir; |
79
|
|
|
|
|
|
|
$output_hash->{data}{mpileup_cmd} = $self->_mpileup_cmd; |
80
|
|
|
|
|
|
|
$output_hash->{data}{max_jobs} = $self->_max_jobs; |
81
|
|
|
|
|
|
|
$output_hash->{data}{fai_chr_regex} = $self->_fai_chr_regex; |
82
|
|
|
|
|
|
|
$output_hash->{data}{fa_ref} = $self->_fa_ref; |
83
|
|
|
|
|
|
|
$output_hash->{data}{fai_ref} = $self->_fai_ref; |
84
|
|
|
|
|
|
|
$output_hash->{data}{ignore_snp_called_status} = $self->_ignore_snp_called_status; |
85
|
|
|
|
|
|
|
$output_hash->{limits} = $self->_escaped_limits; |
86
|
|
|
|
|
|
|
return $output_hash; |
87
|
|
|
|
|
|
|
}; |
88
|
|
|
|
|
|
|
|
89
|
|
|
|
|
|
|
__PACKAGE__->meta->make_immutable; |
90
|
|
|
|
|
|
|
no Moose; |
91
|
|
|
|
|
|
|
1; |
92
|
|
|
|
|
|
|
|
93
|
|
|
|
|
|
|
__END__ |
94
|
|
|
|
|
|
|
|
95
|
|
|
|
|
|
|
=pod |
96
|
|
|
|
|
|
|
|
97
|
|
|
|
|
|
|
=head1 NAME |
98
|
|
|
|
|
|
|
|
99
|
|
|
|
|
|
|
Bio::VertRes::Config::Pipelines::SnpCalling - The base class for the SNP calling pipeline. |
100
|
|
|
|
|
|
|
|
101
|
|
|
|
|
|
|
=head1 VERSION |
102
|
|
|
|
|
|
|
|
103
|
|
|
|
|
|
|
version 1.133090 |
104
|
|
|
|
|
|
|
|
105
|
|
|
|
|
|
|
=head1 SYNOPSIS |
106
|
|
|
|
|
|
|
|
107
|
|
|
|
|
|
|
he base class for the SNP calling pipeline. |
108
|
|
|
|
|
|
|
use Bio::VertRes::Config::Pipelines::SnpCalling; |
109
|
|
|
|
|
|
|
|
110
|
|
|
|
|
|
|
my $pipeline = Bio::VertRes::Config::Pipelines::Mapping->new( |
111
|
|
|
|
|
|
|
database => 'abc', |
112
|
|
|
|
|
|
|
reference => 'Staphylococcus_aureus_subsp_aureus_ABC_v1', |
113
|
|
|
|
|
|
|
limits => { |
114
|
|
|
|
|
|
|
project => ['ABC study'], |
115
|
|
|
|
|
|
|
species => ['EFG'] |
116
|
|
|
|
|
|
|
}, |
117
|
|
|
|
|
|
|
|
118
|
|
|
|
|
|
|
); |
119
|
|
|
|
|
|
|
$pipeline->to_hash(); |
120
|
|
|
|
|
|
|
|
121
|
|
|
|
|
|
|
=head1 AUTHOR |
122
|
|
|
|
|
|
|
|
123
|
|
|
|
|
|
|
Andrew J. Page <ap13@sanger.ac.uk> |
124
|
|
|
|
|
|
|
|
125
|
|
|
|
|
|
|
=head1 COPYRIGHT AND LICENSE |
126
|
|
|
|
|
|
|
|
127
|
|
|
|
|
|
|
This software is Copyright (c) 2013 by Wellcome Trust Sanger Institute. |
128
|
|
|
|
|
|
|
|
129
|
|
|
|
|
|
|
This is free software, licensed under: |
130
|
|
|
|
|
|
|
|
131
|
|
|
|
|
|
|
The GNU General Public License, Version 3, June 2007 |
132
|
|
|
|
|
|
|
|
133
|
|
|
|
|
|
|
=cut |