line |
stmt |
bran |
cond |
sub |
pod |
time |
code |
1
|
|
|
|
|
|
|
undef $VERSION; |
2
|
|
|
|
|
|
|
package Bio::Roary::CommandLine::ExtractProteomeFromGff; |
3
|
|
|
|
|
|
|
$Bio::Roary::CommandLine::ExtractProteomeFromGff::VERSION = '3.11.0'; |
4
|
|
|
|
|
|
|
# ABSTRACT: Take in GFF files and output the proteome |
5
|
|
|
|
|
|
|
|
6
|
|
|
|
|
|
|
|
7
|
1
|
|
|
1
|
|
523743
|
use Moose; |
|
1
|
|
|
|
|
9
|
|
|
1
|
|
|
|
|
8
|
|
8
|
1
|
|
|
1
|
|
7140
|
use Getopt::Long qw(GetOptionsFromArray); |
|
1
|
|
|
|
|
8481
|
|
|
1
|
|
|
|
|
5
|
|
9
|
1
|
|
|
1
|
|
596
|
use Bio::Roary::ExtractProteomeFromGFF; |
|
1
|
|
|
|
|
5
|
|
|
1
|
|
|
|
|
99
|
|
10
|
1
|
|
|
1
|
|
15
|
use File::Basename; |
|
1
|
|
|
|
|
2
|
|
|
1
|
|
|
|
|
604
|
|
11
|
|
|
|
|
|
|
extends 'Bio::Roary::CommandLine::Common'; |
12
|
|
|
|
|
|
|
|
13
|
|
|
|
|
|
|
has 'args' => ( is => 'ro', isa => 'ArrayRef', required => 1 ); |
14
|
|
|
|
|
|
|
has 'script_name' => ( is => 'ro', isa => 'Str', required => 1 ); |
15
|
|
|
|
|
|
|
has 'help' => ( is => 'rw', isa => 'Bool', default => 0 ); |
16
|
|
|
|
|
|
|
|
17
|
|
|
|
|
|
|
has 'gff_files' => ( is => 'rw', isa => 'ArrayRef' ); |
18
|
|
|
|
|
|
|
has 'output_suffix' => ( is => 'rw', isa => 'Str', default => 'proteome.faa' ); |
19
|
|
|
|
|
|
|
has '_error_message' => ( is => 'rw', isa => 'Str' ); |
20
|
|
|
|
|
|
|
has 'apply_unknowns_filter' => ( is => 'rw', isa => 'Bool', default => 1 ); |
21
|
|
|
|
|
|
|
has 'translation_table' => ( is => 'rw', isa => 'Int', default => 11 ); |
22
|
|
|
|
|
|
|
has 'verbose' => ( is => 'rw', isa => 'Bool', default => 0 ); |
23
|
|
|
|
|
|
|
has 'output_directory' => ( is => 'rw', isa => 'Str', default => '.' ); |
24
|
|
|
|
|
|
|
|
25
|
|
|
|
|
|
|
sub BUILD { |
26
|
3
|
|
|
3
|
0
|
14
|
my ($self) = @_; |
27
|
|
|
|
|
|
|
|
28
|
3
|
|
|
|
|
17
|
my ( $gff_files, $output_suffix, $apply_unknowns_filter, $help, $translation_table, $verbose, $cmd_version, $output_directory ); |
29
|
|
|
|
|
|
|
|
30
|
3
|
|
|
|
|
131
|
GetOptionsFromArray( |
31
|
|
|
|
|
|
|
$self->args, |
32
|
|
|
|
|
|
|
'o|output_suffix=s' => \$output_suffix, |
33
|
|
|
|
|
|
|
'f|apply_unknowns_filter=i' => \$apply_unknowns_filter, |
34
|
|
|
|
|
|
|
't|translation_table=i' => \$translation_table, |
35
|
|
|
|
|
|
|
'v|verbose' => \$verbose, |
36
|
|
|
|
|
|
|
'd|output_directory=s' => \$output_directory, |
37
|
|
|
|
|
|
|
'w|version' => \$cmd_version, |
38
|
|
|
|
|
|
|
'h|help' => \$help, |
39
|
|
|
|
|
|
|
); |
40
|
|
|
|
|
|
|
|
41
|
3
|
50
|
|
|
|
2416
|
if ( defined($verbose) ) { |
42
|
0
|
|
|
|
|
0
|
$self->verbose($verbose); |
43
|
0
|
|
|
|
|
0
|
$self->logger->level(10000); |
44
|
|
|
|
|
|
|
} |
45
|
|
|
|
|
|
|
|
46
|
3
|
100
|
|
|
|
46
|
$self->help($help) if(defined($help)); |
47
|
3
|
100
|
|
|
|
108
|
( !$self->help ) or die $self->usage_text; |
48
|
|
|
|
|
|
|
|
49
|
2
|
50
|
|
|
|
11
|
$self->version($cmd_version) if ( defined($cmd_version) ); |
50
|
2
|
50
|
|
|
|
81
|
if ( $self->version ) { |
51
|
0
|
|
|
|
|
0
|
die($self->_version()); |
52
|
|
|
|
|
|
|
} |
53
|
|
|
|
|
|
|
|
54
|
2
|
50
|
|
|
|
8
|
if ( @{ $self->args } == 0 ) { |
|
2
|
|
|
|
|
64
|
|
55
|
0
|
|
|
|
|
0
|
$self->_error_message("Error: You need to provide a GFF file"); |
56
|
|
|
|
|
|
|
} |
57
|
|
|
|
|
|
|
|
58
|
2
|
50
|
|
|
|
10
|
$self->output_suffix($output_suffix) if ( defined($output_suffix) ) ; |
59
|
2
|
50
|
|
|
|
9
|
$self->apply_unknowns_filter($apply_unknowns_filter) if ( defined($apply_unknowns_filter) ); |
60
|
2
|
100
|
|
|
|
38
|
$self->translation_table($translation_table) if ( defined($translation_table) ); |
61
|
2
|
50
|
|
|
|
7
|
$self->output_directory($output_directory) if ( defined($output_directory) ); |
62
|
|
|
|
|
|
|
|
63
|
2
|
|
|
|
|
6
|
for my $filename ( @{ $self->args } ) { |
|
2
|
|
|
|
|
54
|
|
64
|
2
|
50
|
|
|
|
86
|
if ( !-e $filename ) { |
65
|
0
|
|
|
|
|
0
|
$self->_error_message("Error: Cant access file $filename"); |
66
|
0
|
|
|
|
|
0
|
last; |
67
|
|
|
|
|
|
|
} |
68
|
|
|
|
|
|
|
} |
69
|
2
|
|
|
|
|
56
|
$self->gff_files( $self->args ); |
70
|
|
|
|
|
|
|
|
71
|
|
|
|
|
|
|
} |
72
|
|
|
|
|
|
|
|
73
|
|
|
|
|
|
|
sub run { |
74
|
2
|
|
|
2
|
0
|
6
|
my ($self) = @_; |
75
|
|
|
|
|
|
|
|
76
|
2
|
50
|
|
|
|
53
|
if ( defined( $self->_error_message ) ) { |
77
|
0
|
|
|
|
|
0
|
print $self->_error_message . "\n"; |
78
|
0
|
|
|
|
|
0
|
die $self->usage_text; |
79
|
|
|
|
|
|
|
} |
80
|
|
|
|
|
|
|
|
81
|
2
|
|
|
|
|
6
|
for my $gff_file ( @{ $self->gff_files } ) { |
|
2
|
|
|
|
|
45
|
|
82
|
2
|
|
|
|
|
84
|
my ( $filename, $directories, $suffix ) = fileparse($gff_file); |
83
|
2
|
|
|
|
|
61
|
my $obj = Bio::Roary::ExtractProteomeFromGFF->new( |
84
|
|
|
|
|
|
|
gff_file => $gff_file, |
85
|
|
|
|
|
|
|
output_filename => $filename . '.' . $self->output_suffix, |
86
|
|
|
|
|
|
|
apply_unknowns_filter => $self->apply_unknowns_filter, |
87
|
|
|
|
|
|
|
translation_table => $self->translation_table, |
88
|
|
|
|
|
|
|
output_directory => $self->output_directory, |
89
|
|
|
|
|
|
|
); |
90
|
2
|
|
|
|
|
61
|
$obj->fasta_file(); |
91
|
|
|
|
|
|
|
} |
92
|
|
|
|
|
|
|
|
93
|
|
|
|
|
|
|
} |
94
|
|
|
|
|
|
|
|
95
|
|
|
|
|
|
|
sub usage_text { |
96
|
1
|
|
|
1
|
0
|
4
|
my ($self) = @_; |
97
|
|
|
|
|
|
|
|
98
|
1
|
|
|
|
|
35
|
return <<USAGE; |
99
|
|
|
|
|
|
|
Usage: extract_proteome_from_gff [options] *.gff |
100
|
|
|
|
|
|
|
Take in GFF files and create FASTA files of the protein sequences |
101
|
|
|
|
|
|
|
|
102
|
|
|
|
|
|
|
Options: -o STR output suffix [proteome.faa] |
103
|
|
|
|
|
|
|
-t INT translation table [11] |
104
|
|
|
|
|
|
|
-f filter sequences with missing data |
105
|
|
|
|
|
|
|
-v verbose output to STDOUT |
106
|
|
|
|
|
|
|
-d STR output directory |
107
|
|
|
|
|
|
|
-w print version and exit |
108
|
|
|
|
|
|
|
-h this help message |
109
|
|
|
|
|
|
|
|
110
|
|
|
|
|
|
|
For further info see: http://sanger-pathogens.github.io/Roary/ |
111
|
|
|
|
|
|
|
USAGE |
112
|
|
|
|
|
|
|
} |
113
|
|
|
|
|
|
|
|
114
|
|
|
|
|
|
|
__PACKAGE__->meta->make_immutable; |
115
|
1
|
|
|
1
|
|
7
|
no Moose; |
|
1
|
|
|
|
|
2
|
|
|
1
|
|
|
|
|
10
|
|
116
|
|
|
|
|
|
|
1; |
117
|
|
|
|
|
|
|
|
118
|
|
|
|
|
|
|
__END__ |
119
|
|
|
|
|
|
|
|
120
|
|
|
|
|
|
|
=pod |
121
|
|
|
|
|
|
|
|
122
|
|
|
|
|
|
|
=encoding UTF-8 |
123
|
|
|
|
|
|
|
|
124
|
|
|
|
|
|
|
=head1 NAME |
125
|
|
|
|
|
|
|
|
126
|
|
|
|
|
|
|
Bio::Roary::CommandLine::ExtractProteomeFromGff - Take in GFF files and output the proteome |
127
|
|
|
|
|
|
|
|
128
|
|
|
|
|
|
|
=head1 VERSION |
129
|
|
|
|
|
|
|
|
130
|
|
|
|
|
|
|
version 3.11.0 |
131
|
|
|
|
|
|
|
|
132
|
|
|
|
|
|
|
=head1 SYNOPSIS |
133
|
|
|
|
|
|
|
|
134
|
|
|
|
|
|
|
Take in a GFF file and output the proteome |
135
|
|
|
|
|
|
|
|
136
|
|
|
|
|
|
|
=head1 AUTHOR |
137
|
|
|
|
|
|
|
|
138
|
|
|
|
|
|
|
Andrew J. Page <ap13@sanger.ac.uk> |
139
|
|
|
|
|
|
|
|
140
|
|
|
|
|
|
|
=head1 COPYRIGHT AND LICENSE |
141
|
|
|
|
|
|
|
|
142
|
|
|
|
|
|
|
This software is Copyright (c) 2013 by Wellcome Trust Sanger Institute. |
143
|
|
|
|
|
|
|
|
144
|
|
|
|
|
|
|
This is free software, licensed under: |
145
|
|
|
|
|
|
|
|
146
|
|
|
|
|
|
|
The GNU General Public License, Version 3, June 2007 |
147
|
|
|
|
|
|
|
|
148
|
|
|
|
|
|
|
=cut |