line |
stmt |
bran |
cond |
sub |
pod |
time |
code |
1
|
|
|
|
|
|
|
package Bio::MLST::ProcessFasta; |
2
|
|
|
|
|
|
|
# ABSTRACT: Take in a fasta file, lookup the MLST database and create relevant files. |
3
|
|
|
|
|
|
|
$Bio::MLST::ProcessFasta::VERSION = '2.1.1706216'; |
4
|
|
|
|
|
|
|
|
5
|
|
|
|
|
|
|
|
6
|
9
|
|
|
9
|
|
34
|
use Moose; |
|
9
|
|
|
|
|
11
|
|
|
9
|
|
|
|
|
40
|
|
7
|
9
|
|
|
9
|
|
42358
|
use Bio::MLST::SearchForFiles; |
|
9
|
|
|
|
|
20
|
|
|
9
|
|
|
|
|
302
|
|
8
|
9
|
|
|
9
|
|
4291
|
use Bio::MLST::CompareAlleles; |
|
9
|
|
|
|
|
18
|
|
|
9
|
|
|
|
|
325
|
|
9
|
9
|
|
|
9
|
|
67
|
use Bio::MLST::SequenceType; |
|
9
|
|
|
|
|
11
|
|
|
9
|
|
|
|
|
142
|
|
10
|
9
|
|
|
9
|
|
4230
|
use Bio::MLST::OutputFasta; |
|
9
|
|
|
|
|
25
|
|
|
9
|
|
|
|
|
329
|
|
11
|
9
|
|
|
9
|
|
4624
|
use Bio::MLST::Spreadsheet::Row; |
|
9
|
|
|
|
|
20
|
|
|
9
|
|
|
|
|
332
|
|
12
|
9
|
|
|
9
|
|
62
|
use Bio::MLST::Types; |
|
9
|
|
|
|
|
16
|
|
|
9
|
|
|
|
|
2776
|
|
13
|
|
|
|
|
|
|
|
14
|
|
|
|
|
|
|
has 'species' => ( is => 'ro', isa => 'Str', required => 1 ); |
15
|
|
|
|
|
|
|
has 'base_directory' => ( is => 'ro', isa => 'Str', required => 1 ); |
16
|
|
|
|
|
|
|
has 'fasta_file' => ( is => 'ro', isa => 'Bio::MLST::File', required => 1 ); |
17
|
|
|
|
|
|
|
has 'makeblastdb_exec' => ( is => 'ro', isa => 'Str', required => 1 ); |
18
|
|
|
|
|
|
|
has 'blastn_exec' => ( is => 'ro', isa => 'Str', required => 1 ); |
19
|
|
|
|
|
|
|
has 'output_directory' => ( is => 'ro', isa => 'Str', required => 1 ); |
20
|
|
|
|
|
|
|
has 'output_fasta_files' => ( is => 'ro', isa => 'Bool', default => 0 ); |
21
|
|
|
|
|
|
|
has 'show_contamination_instead_of_alt_matches' => ( is => 'ro', isa => 'Bool', default => 1 ); |
22
|
|
|
|
|
|
|
has 'report_lowest_st' => ( is => 'ro', isa => 'Bool', default => 0 ); |
23
|
|
|
|
|
|
|
|
24
|
|
|
|
|
|
|
has '_search_results' => ( is => 'ro', isa => 'Bio::MLST::SearchForFiles', lazy => 1, builder => '_build__search_results' ); |
25
|
|
|
|
|
|
|
has '_compare_alleles' => ( is => 'ro', isa => 'Bio::MLST::CompareAlleles', lazy => 1, builder => '_build__compare_alleles' ); |
26
|
|
|
|
|
|
|
has '_sequence_type_obj' => ( is => 'ro', isa => 'Bio::MLST::SequenceType', lazy => 1, builder => '_build__sequence_type_obj' ); |
27
|
|
|
|
|
|
|
has '_spreadsheet_row_obj' => ( is => 'ro', isa => 'Bio::MLST::Spreadsheet::Row', lazy => 1, builder => '_build__spreadsheet_row_obj' ); |
28
|
|
|
|
|
|
|
|
29
|
|
|
|
|
|
|
has 'concat_name' => ( is => 'rw', isa => 'Maybe[Str]' ); |
30
|
|
|
|
|
|
|
has 'concat_sequence' => ( is => 'rw', isa => 'Maybe[Str]' ); |
31
|
|
|
|
|
|
|
|
32
|
|
|
|
|
|
|
sub _build__search_results |
33
|
|
|
|
|
|
|
{ |
34
|
4
|
|
|
4
|
|
25
|
my($self) = @_; |
35
|
4
|
|
|
|
|
149
|
Bio::MLST::SearchForFiles->new( |
36
|
|
|
|
|
|
|
species_name => $self->species, |
37
|
|
|
|
|
|
|
base_directory => $self->base_directory |
38
|
|
|
|
|
|
|
); |
39
|
|
|
|
|
|
|
} |
40
|
|
|
|
|
|
|
|
41
|
|
|
|
|
|
|
sub _build__compare_alleles |
42
|
|
|
|
|
|
|
{ |
43
|
4
|
|
|
4
|
|
9
|
my($self) = @_; |
44
|
4
|
|
|
|
|
124
|
my $compare_alleles = Bio::MLST::CompareAlleles->new( |
45
|
|
|
|
|
|
|
sequence_filename => $self->fasta_file, |
46
|
|
|
|
|
|
|
allele_filenames => $self->_search_results->allele_filenames(), |
47
|
|
|
|
|
|
|
makeblastdb_exec => $self->makeblastdb_exec, |
48
|
|
|
|
|
|
|
blastn_exec => $self->blastn_exec, |
49
|
|
|
|
|
|
|
profiles_filename => $self->_search_results->profiles_filename() |
50
|
|
|
|
|
|
|
); |
51
|
|
|
|
|
|
|
|
52
|
4
|
|
|
|
|
152
|
my $output_fasta = Bio::MLST::OutputFasta->new( |
53
|
|
|
|
|
|
|
matching_sequences => $compare_alleles->matching_sequences, |
54
|
|
|
|
|
|
|
non_matching_sequences => $compare_alleles->non_matching_sequences, |
55
|
|
|
|
|
|
|
output_directory => $self->output_directory, |
56
|
|
|
|
|
|
|
input_fasta_file => $self->fasta_file |
57
|
|
|
|
|
|
|
); |
58
|
0
|
|
|
|
|
0
|
$output_fasta->create_files(); |
59
|
0
|
|
|
|
|
0
|
$self->concat_name($output_fasta->_fasta_filename); |
60
|
0
|
|
|
|
|
0
|
$self->concat_sequence($output_fasta->concat_sequence); |
61
|
0
|
|
|
|
|
0
|
return $compare_alleles; |
62
|
|
|
|
|
|
|
} |
63
|
|
|
|
|
|
|
|
64
|
|
|
|
|
|
|
sub _build__sequence_type_obj |
65
|
|
|
|
|
|
|
{ |
66
|
4
|
|
|
4
|
|
11
|
my($self) = @_; |
67
|
4
|
|
|
|
|
100
|
my $sequence_type_obj = Bio::MLST::SequenceType->new( |
68
|
|
|
|
|
|
|
profiles_filename => $self->_search_results->profiles_filename(), |
69
|
|
|
|
|
|
|
matching_names => $self->_compare_alleles->found_sequence_names, |
70
|
|
|
|
|
|
|
non_matching_names => $self->_compare_alleles->found_non_matching_sequence_names, |
71
|
|
|
|
|
|
|
report_lowest_st => $self->report_lowest_st |
72
|
|
|
|
|
|
|
); |
73
|
|
|
|
|
|
|
} |
74
|
|
|
|
|
|
|
|
75
|
|
|
|
|
|
|
sub _build__spreadsheet_row_obj |
76
|
|
|
|
|
|
|
{ |
77
|
4
|
|
|
4
|
|
17
|
my($self) = @_; |
78
|
4
|
|
|
|
|
147
|
Bio::MLST::Spreadsheet::Row->new( |
79
|
|
|
|
|
|
|
sequence_type_obj => $self->_sequence_type_obj, |
80
|
|
|
|
|
|
|
compare_alleles => $self->_compare_alleles, |
81
|
|
|
|
|
|
|
show_contamination_instead_of_alt_matches => $self->show_contamination_instead_of_alt_matches |
82
|
|
|
|
|
|
|
); |
83
|
|
|
|
|
|
|
} |
84
|
|
|
|
|
|
|
|
85
|
9
|
|
|
9
|
|
45
|
no Moose; |
|
9
|
|
|
|
|
13
|
|
|
9
|
|
|
|
|
50
|
|
86
|
|
|
|
|
|
|
__PACKAGE__->meta->make_immutable; |
87
|
|
|
|
|
|
|
1; |
88
|
|
|
|
|
|
|
|
89
|
|
|
|
|
|
|
__END__ |
90
|
|
|
|
|
|
|
|
91
|
|
|
|
|
|
|
=pod |
92
|
|
|
|
|
|
|
|
93
|
|
|
|
|
|
|
=encoding UTF-8 |
94
|
|
|
|
|
|
|
|
95
|
|
|
|
|
|
|
=head1 NAME |
96
|
|
|
|
|
|
|
|
97
|
|
|
|
|
|
|
Bio::MLST::ProcessFasta - Take in a fasta file, lookup the MLST database and create relevant files. |
98
|
|
|
|
|
|
|
|
99
|
|
|
|
|
|
|
=head1 VERSION |
100
|
|
|
|
|
|
|
|
101
|
|
|
|
|
|
|
version 2.1.1706216 |
102
|
|
|
|
|
|
|
|
103
|
|
|
|
|
|
|
=head1 SYNOPSIS |
104
|
|
|
|
|
|
|
|
105
|
|
|
|
|
|
|
Take in a fasta file, lookup the MLST database and create relevant files. |
106
|
|
|
|
|
|
|
|
107
|
|
|
|
|
|
|
use Bio::MLST::ProcessFasta; |
108
|
|
|
|
|
|
|
Bio::MLST::ProcessFasta->new( |
109
|
|
|
|
|
|
|
'species' => 'E.coli', |
110
|
|
|
|
|
|
|
'base_directory' => '/path/to/dir', |
111
|
|
|
|
|
|
|
'fasta_file' => 'myfasta.fa', |
112
|
|
|
|
|
|
|
'makeblastdb_exec' => 'makeblastdb', |
113
|
|
|
|
|
|
|
'blastn_exec' => 'blastn', |
114
|
|
|
|
|
|
|
'output_directory' => '/path/to/output', |
115
|
|
|
|
|
|
|
'output_fasta_files'=> 1, |
116
|
|
|
|
|
|
|
); |
117
|
|
|
|
|
|
|
|
118
|
|
|
|
|
|
|
=head1 METHODS |
119
|
|
|
|
|
|
|
|
120
|
|
|
|
|
|
|
=head2 concat_name |
121
|
|
|
|
|
|
|
|
122
|
|
|
|
|
|
|
Output the name of the concatinated multifasta file |
123
|
|
|
|
|
|
|
|
124
|
|
|
|
|
|
|
=head2 concat_sequence |
125
|
|
|
|
|
|
|
|
126
|
|
|
|
|
|
|
Output the sequences of the concatinated multifasta file |
127
|
|
|
|
|
|
|
|
128
|
|
|
|
|
|
|
=head2 _spreadsheet_row_obj |
129
|
|
|
|
|
|
|
|
130
|
|
|
|
|
|
|
A row in the output spreadsheet |
131
|
|
|
|
|
|
|
|
132
|
|
|
|
|
|
|
=head1 SEE ALSO |
133
|
|
|
|
|
|
|
|
134
|
|
|
|
|
|
|
=over 4 |
135
|
|
|
|
|
|
|
|
136
|
|
|
|
|
|
|
=item * |
137
|
|
|
|
|
|
|
|
138
|
|
|
|
|
|
|
L<Bio::MLST::Check> |
139
|
|
|
|
|
|
|
|
140
|
|
|
|
|
|
|
=back |
141
|
|
|
|
|
|
|
|
142
|
|
|
|
|
|
|
=head1 AUTHOR |
143
|
|
|
|
|
|
|
|
144
|
|
|
|
|
|
|
Andrew J. Page <ap13@sanger.ac.uk> |
145
|
|
|
|
|
|
|
|
146
|
|
|
|
|
|
|
=head1 COPYRIGHT AND LICENSE |
147
|
|
|
|
|
|
|
|
148
|
|
|
|
|
|
|
This software is Copyright (c) 2012 by Wellcome Trust Sanger Institute. |
149
|
|
|
|
|
|
|
|
150
|
|
|
|
|
|
|
This is free software, licensed under: |
151
|
|
|
|
|
|
|
|
152
|
|
|
|
|
|
|
The GNU General Public License, Version 3, June 2007 |
153
|
|
|
|
|
|
|
|
154
|
|
|
|
|
|
|
=cut |