0
|
1 undef $VERSION;
|
|
2 package Bio::Roary::CommandLine::IterativeCdhit;
|
|
3
|
|
4 # ABSTRACT: Iteratively run cdhit
|
|
5
|
|
6 =head1 SYNOPSIS
|
|
7
|
|
8 Iteratively run cdhit
|
|
9
|
|
10 =cut
|
|
11
|
|
12 use Moose;
|
|
13 use Getopt::Long qw(GetOptionsFromArray);
|
|
14 use Bio::Roary::IterativeCdhit;
|
|
15 extends 'Bio::Roary::CommandLine::Common';
|
|
16
|
|
17 has 'args' => ( is => 'ro', isa => 'ArrayRef', required => 1 );
|
|
18 has 'script_name' => ( is => 'ro', isa => 'Str', required => 1 );
|
|
19 has 'help' => ( is => 'rw', isa => 'Bool', default => 0 );
|
|
20 has '_error_message' => ( is => 'rw', isa => 'Str' );
|
|
21
|
|
22 has 'output_cd_hit_filename' => ( is => 'rw', isa => 'Str', default => '_clustered' );
|
|
23 has 'output_combined_filename' => ( is => 'rw', isa => 'Str', default => '_combined_files' );
|
|
24 has 'number_of_input_files' => ( is => 'rw', isa => 'Int', default => 1 );
|
|
25 has 'output_filtered_clustered_fasta' => ( is => 'rw', isa => 'Str', default => '_clustered_filtered.fa' );
|
|
26
|
|
27 has 'lower_bound_percentage' => ( is => 'rw', isa => 'Num', default => 0.98 );
|
|
28 has 'upper_bound_percentage' => ( is => 'rw', isa => 'Num', default => 0.99 );
|
|
29 has 'step_size_percentage' => ( is => 'rw', isa => 'Num', default => 0.005 );
|
|
30 has 'cpus' => ( is => 'rw', isa => 'Int', default => 1 );
|
|
31 has 'verbose' => ( is => 'rw', isa => 'Bool', default => 0 );
|
|
32
|
|
33
|
|
34 sub BUILD {
|
|
35 my ($self) = @_;
|
|
36
|
|
37 my ( $output_cd_hit_filename,$cpus,$lower_bound_percentage,$upper_bound_percentage,$step_size_percentage, $output_combined_filename, $number_of_input_files, $output_filtered_clustered_fasta,$verbose,
|
|
38 $help );
|
|
39
|
|
40 GetOptionsFromArray(
|
|
41 $self->args,
|
|
42 'c|output_cd_hit_filename=s' => \$output_cd_hit_filename,
|
|
43 'm|output_combined_filename=s' => \$output_combined_filename,
|
|
44 'n|number_of_input_files=i' => \$number_of_input_files,
|
|
45 'f|output_filtered_clustered_fasta=s' => \$output_filtered_clustered_fasta,
|
|
46 'l|lower_bound_percentage=s' => \$lower_bound_percentage,
|
|
47 'u|upper_bound_percentage=s' => \$upper_bound_percentage,
|
|
48 's|step_size_percentage=s' => \$step_size_percentage,
|
|
49 'p|cpus=i' => \$cpus,
|
|
50 'v|verbose' => \$verbose,
|
|
51 'h|help' => \$help,
|
|
52 );
|
|
53
|
|
54 if ( defined($verbose) ) {
|
|
55 $self->verbose($verbose);
|
|
56 $self->logger->level(10000);
|
|
57 }
|
|
58 $self->help($help) if(defined($help));
|
|
59 $self->lower_bound_percentage($lower_bound_percentage/100) if ( defined($lower_bound_percentage) );
|
|
60 $self->upper_bound_percentage($upper_bound_percentage/100) if ( defined($upper_bound_percentage) );
|
|
61 $self->step_size_percentage($step_size_percentage/100) if ( defined($step_size_percentage) );
|
|
62 $self->output_cd_hit_filename($output_cd_hit_filename) if ( defined($output_cd_hit_filename) );
|
|
63 $self->output_combined_filename($output_combined_filename) if ( defined($output_combined_filename) );
|
|
64 $self->number_of_input_files($number_of_input_files) if ( defined($number_of_input_files) );
|
|
65 $self->cpus($cpus) if ( defined($cpus) );
|
|
66 $self->output_filtered_clustered_fasta($output_filtered_clustered_fasta)
|
|
67 if ( defined($output_filtered_clustered_fasta) );
|
|
68
|
|
69 }
|
|
70
|
|
71 sub run {
|
|
72 my ($self) = @_;
|
|
73
|
|
74 ( !$self->help ) or die $self->usage_text;
|
|
75 if ( defined( $self->_error_message ) ) {
|
|
76 print $self->_error_message . "\n";
|
|
77 die $self->usage_text;
|
|
78 }
|
|
79
|
|
80 my $obj = Bio::Roary::IterativeCdhit->new(
|
|
81 output_cd_hit_filename => $self->output_cd_hit_filename,
|
|
82 output_combined_filename => $self->output_combined_filename,
|
|
83 number_of_input_files => $self->number_of_input_files,
|
|
84 output_filtered_clustered_fasta => $self->output_filtered_clustered_fasta,
|
|
85 lower_bound_percentage => $self->lower_bound_percentage,
|
|
86 upper_bound_percentage => $self->upper_bound_percentage,
|
|
87 step_size_percentage => $self->step_size_percentage,
|
|
88 cpus => $self->cpus,
|
|
89 logger => $self->logger
|
|
90
|
|
91 );
|
|
92 $obj->run;
|
|
93 }
|
|
94
|
|
95 sub usage_text {
|
|
96 my ($self) = @_;
|
|
97
|
|
98 return <<USAGE;
|
|
99 Usage: iterative_cdhit [options]
|
|
100 Iteratively cluster a FASTA file of proteins with CD-hit, lower the threshold each time and extracting core genes (1 per isolate) to another file, and remove them from the input proteins file.
|
|
101
|
|
102 Required arguments:
|
|
103 -m STR input FASTA file of protein sequences [_combined_files]
|
|
104
|
|
105 Options: -p INT number of threads [1]
|
|
106 -n INT number of isolates [1]
|
|
107 -c STR cd-hit output filename [_clustered]
|
|
108 -f STR output filename for filtered sequences [_clustered_filtered.fa]
|
|
109 -l FLOAT lower bound percentage identity [98.0]
|
|
110 -u FLOAT upper bound percentage identity [99.0]
|
|
111 -s FLOAT step size for percentage identity [0.5]
|
|
112 -v verbose output to STDOUT
|
|
113 -h this help message
|
|
114
|
|
115 For further info see: http://sanger-pathogens.github.io/Roary/
|
|
116 USAGE
|
|
117 }
|
|
118
|
|
119 __PACKAGE__->meta->make_immutable;
|
|
120 no Moose;
|
|
121 1;
|