comparison Roary/lib/Bio/Roary/ChunkFastaFile.pm @ 0:c47a5f61bc9f draft

Uploaded
author dereeper
date Fri, 14 May 2021 20:27:06 +0000
parents
children
comparison
equal deleted inserted replaced
-1:000000000000 0:c47a5f61bc9f
1 package Bio::Roary::ChunkFastaFile;
2
3 # ABSTRACT: Take in a FASTA file and chunk it up into smaller pieces.
4
5 =head1 SYNOPSIS
6
7 Take in a FASTA file and chunk it up into smaller pieces.
8 use Bio::Roary::ChunkFastaFile;
9
10 my $obj = Bio::Roary::ChunkFastaFile->new(
11 fasta_file => 'abc.fa',
12 );
13 $obj->sequence_file_names;
14
15 =cut
16
17 use Moose;
18 use Bio::SeqIO;
19 use Bio::Roary::Exceptions;
20 use Cwd;
21 use File::Temp;
22
23 has 'fasta_file' => ( is => 'ro', isa => 'Str', required => 1 );
24 has 'target_chunk_size' => ( is => 'ro', isa => 'Int', default => 200000 );
25 has 'sequence_file_names' => ( is => 'ro', isa => 'ArrayRef', lazy => 1, builder => '_build_sequence_file_names' );
26 has '_working_directory' =>
27 ( is => 'ro', isa => 'File::Temp::Dir', default => sub { File::Temp->newdir( DIR => getcwd, CLEANUP => 1 ); } );
28 has '_working_directory_name' => ( is => 'ro', isa => 'Str', lazy => 1, builder => '_build__working_directory_name' );
29 has '_input_seqio' => ( is => 'ro', isa => 'Bio::SeqIO', lazy => 1, builder => '_build__input_seqio' );
30
31 sub _build__working_directory_name {
32 my ($self) = @_;
33 return $self->_working_directory->dirname();
34 }
35
36 sub _build__input_seqio {
37 my ($self) = @_;
38 return Bio::SeqIO->new( -file => $self->fasta_file, -format => 'Fasta' );
39 }
40
41 sub _create_next_chunk_file_name {
42 my ( $self, $chunk_number ) = @_;
43 return join( '/', ( $self->_working_directory_name, $chunk_number . '.seq' ) );
44 }
45
46 sub _create_next_chunk_seqio {
47 my ( $self, $chunk_number ) = @_;
48 return Bio::SeqIO->new( -file => ">".$self->_create_next_chunk_file_name($chunk_number), -format => 'Fasta' );
49 }
50
51 sub _build_sequence_file_names {
52 my ($self) = @_;
53 my @sequence_file_names;
54 my $chunk_number = 0;
55 my $current_chunk_length = 0;
56 my $current_chunk_seqio = $self->_create_next_chunk_seqio($chunk_number);
57 push( @sequence_file_names, $self->_create_next_chunk_file_name($chunk_number) );
58
59 while ( my $input_seq = $self->_input_seqio->next_seq() ) {
60 if ( $current_chunk_length > $self->target_chunk_size ) {
61
62 # next chunk
63 $chunk_number++;
64 $current_chunk_length = 0;
65 $current_chunk_seqio = $self->_create_next_chunk_seqio($chunk_number);
66 push( @sequence_file_names, $self->_create_next_chunk_file_name($chunk_number) );
67 }
68 $current_chunk_seqio->write_seq($input_seq);
69 $current_chunk_length += $input_seq->length();
70 }
71 return \@sequence_file_names;
72 }
73
74 no Moose;
75 __PACKAGE__->meta->make_immutable;
76
77 1;