Skip to content

Commit

Permalink
Add tempdir. Prep for maintenance release
Browse files Browse the repository at this point in the history
  • Loading branch information
@Ebedthan
Ebedthan committed Mar 25, 2021
1 parent ae96d21 commit 987d600
Showing 1 changed file with 38 additions and 33 deletions.
71 changes: 38 additions & 33 deletions itap
View file
Original file line number Diff line number Diff line change
Expand Up @@ -14,7 +14,7 @@ use File::Copy;
use File::Log;
use File::Path qw( remove_tree );
use File::Spec;
use File::Temp qw( tempfile );
use File::Temp qw( tempfile tempdir );
use FindBin qw( $RealBin );
use Getopt::Declare;
use IPC::System::Simple qw( capture );
Expand All @@ -26,13 +26,13 @@ use Text::CSV::Hashify;
use Time::Piece;
use Time::Seconds;

our $VERSION = '0.1.0';
our $VERSION = '0.1.1';

# Global variables
my @CMDLINE = ( $PROGRAM_NAME, @ARGV );
my $BINDIR = "$FindBin::RealBin/../binaries/$OSNAME";
my $EXE = $FindBin::RealScript;
my $AUTHOR = 'Anicet Ebou <anicet.ebou@gmail.com>';
my $AUTHOR = 'Anicet Ebou <anicet.ebou@gmail.com> and Dominique Koua.';
my $URL = 'https://github.com/koualab/itap.git';
my $EMPTY = q{};
my $SPACE = q{ };
Expand Down Expand Up @@ -134,6 +134,9 @@ my $log = File::Log->new(
logFileMode => '>',
);

# Setup temp dir
my $tmpdir = tempdir( CLEANUP => 1);

# START ----------------------------------------------------------------------
msg( 'This is ' . $EXE . $SPACE . $VERSION );
msg( 'Written by ' . $AUTHOR );
Expand Down Expand Up @@ -178,16 +181,16 @@ if ( index( $signalp_version, 'version 5' ) == $NON_EXIST ) {
msg('Builing transcripts index');
msg("Use $kallisto_version");
runcmd( 'kallisto index ' . '-i '
. File::Spec->catfile( $outdir, 'all_transcripts.index' )
. File::Spec->catfile( $tmpdir, 'all_transcripts.index' )
. " $transcripts" );
msg('Done creating index');

# Quantifying reads
msg('Quantifying reads');
runcmd( 'kallisto quant ' . '-t '
. $cpus . ' -i '
. File::Spec->catfile( $outdir, 'all_transcripts.index' ) . ' -o '
. File::Spec->catfile( $outdir, 'kallisto_1' )
. File::Spec->catfile( $tmpdir, 'all_transcripts.index' ) . ' -o '
. File::Spec->catfile( $tmpdir, 'kallisto_1' )
. " $forward_reads $reverse_reads" );
msg('Done quantifying reads');

Expand All @@ -196,13 +199,14 @@ run_translation($transcripts);

# STEP 3: Putative toxins identification--------------------------------------
msg('Putative toxins prediction');

runcmd( 'hmmcompete --desc --pepreg '
. '--hmm '
. File::Spec->catfile( $dbdir, 'ekenda.hmm' )
. ' --in '
. File::Spec->catfile( $outdir, 'contigs_prot.fas' )
. File::Spec->catfile( $tmpdir, 'contigs_prot.fas' )
. ' --out '
. File::Spec->catfile( $outdir, 'transcripts_hmmcompete_out.txt' ) );
. File::Spec->catfile( $tmpdir, 'transcripts_hmmcompete_out.txt' ) );

# Retrieve DNA sequences of AA contigs having a match in hmmcompete

Expand All @@ -211,12 +215,12 @@ runcmd( 'hmmcompete --desc --pepreg '
msg('Adding description of matched families');
my %fields
= add_desc_to_ids(
File::Spec->catfile( $outdir, 'transcripts_hmmcompete_out.txt' ) );
File::Spec->catfile( $tmpdir, 'transcripts_hmmcompete_out.txt' ) );

## Second, get DNA sequences using ids of putatives sequences
## from hmmcompete result.
msg('Getting DNA sequences using ids of putatives sequences');
my $contigs_dna_file = File::Spec->catfile( $outdir, 'contigs_dna.fas' );
my $contigs_dna_file = File::Spec->catfile( $tmpdir, 'contigs_dna.fas' );
open my $precdna, '>', $contigs_dna_file
or croak "Opening $contigs_dna_file file failed: $OS_ERROR";
get_dna_seq( $transcripts, $precdna, \%fields );
Expand All @@ -226,10 +230,10 @@ close $precdna or croak "Close $precdna failed: $OS_ERROR";
# Extract matching region from hmmcompete.out and extend sequence
msg('Extracting matching region and extending sequence');
my $contigs_protreg_file
= File::Spec->catfile( $outdir, 'contigs_prot_reg.fas' );
= File::Spec->catfile( $tmpdir, 'contigs_prot_reg.fas' );
open my $extseq, '>', $contigs_protreg_file
or croak "Open $contigs_protreg_file file failed: $OS_ERROR";
extract_seq( \%fields, File::Spec->catfile( $outdir, 'contigs_prot.fas' ),
extract_seq( \%fields, File::Spec->catfile( $tmpdir, 'contigs_prot.fas' ),
$extseq );
close $extseq or croak "Close $extseq failed: $OS_ERROR";

Expand All @@ -248,31 +252,31 @@ runcmd( 'signalp '
. '-gff3' );
move(
File::Spec->catfile( $FindBin::RealBin, 'contigs_prot_reg.gff3' ),
File::Spec->catfile( $outdir, 'contigs_prot_reg.gff3' )
File::Spec->catfile( $tmpdir, 'contigs_prot_reg.gff3' )
);
move(
File::Spec->catfile(
$FindBin::RealBin, 'contigs_prot_reg_summary.signalp5'
),
File::Spec->catfile( $outdir, 'contigs_prot_reg_summary.signalp5' )
File::Spec->catfile( $tmpdir, 'contigs_prot_reg_summary.signalp5' )
);

msg('Done searching signal sequences');

mkdir File::Spec->catfile( $outdir, 'famaln' ), $RWE_FOR_OWNER;
mkdir File::Spec->catfile( $tmpdir, 'famaln' ), $RWE_FOR_OWNER;
split_by_fam($contigs_protreg_file);

msg('Adding signal sequence presence to sequence header');
add_signal_presence( File::Spec->catfile( $outdir, 'contigs_prot_reg.gff3' ),
add_signal_presence( File::Spec->catfile( $tmpdir, 'contigs_prot_reg.gff3' ),
$contigs_protreg_file );
msg('Done adding signal presence');

# Get DNA region
process_get_dna_reg(
$transcripts,
File::Spec->catfile( $outdir, 'contigs_prot.fas' ),
File::Spec->catfile( $outdir, 'contigs_prot_reg_new.fas' ),
File::Spec->catfile( $outdir, 'contigs_dna_reg.fas' )
File::Spec->catfile( $tmpdir, 'contigs_prot.fas' ),
File::Spec->catfile( $tmpdir, 'contigs_prot_reg_new.fas' ),
File::Spec->catfile( $tmpdir, 'contigs_dna_reg.fas' )
);

# STEP 7: Compute TPM for precursors------------------------------------------
Expand All @@ -281,29 +285,29 @@ msg('Computing expression level for precursors');
# Build new index from putative transcripts
msg('Building putative transcripts index');
runcmd( 'kallisto index' . ' -i '
. File::Spec->catfile( $outdir, 'putative_transcripts.index' )
. File::Spec->catfile( $tmpdir, 'putative_transcripts.index' )
. $SPACE
. File::Spec->catfile( $outdir, 'contigs_dna_reg.fas' ) );
. File::Spec->catfile( $tmpdir, 'contigs_dna_reg.fas' ) );
msg('Done building putative transcripts index');

# Quantify raw reads using putative transcripts index
msg('Quantifying putative transcripts');
runcmd( 'kallisto quant ' . '-t '
. $cpus . ' -i '
. File::Spec->catfile( $outdir, 'putative_transcripts.index' )
. File::Spec->catfile( $tmpdir, 'putative_transcripts.index' )
. ' -o '
. File::Spec->catdir( $outdir, 'kallisto_2' )
. File::Spec->catdir( $tmpdir, 'kallisto_2' )
. " $forward_reads $reverse_reads" );
msg('Done computing expression level for precursors');

# STEP 8: Writing output stats------------------------------------------------
msg('Writing output stats');

my $kh1_ref
= hashify( File::Spec->catfile( $outdir, 'kallisto_1', 'abundance.tsv' ),
= hashify( File::Spec->catfile( $tmpdir, 'kallisto_1', 'abundance.tsv' ),
'target_id' );
my $kh2_ref
= hashify( File::Spec->catfile( $outdir, 'kallisto_2', 'abundance.tsv' ),
= hashify( File::Spec->catfile( $tmpdir, 'kallisto_2', 'abundance.tsv' ),
'target_id' );

my $itap_outfile = File::Spec->catfile( $outdir, 'itap_output.tsv' );
Expand Down Expand Up @@ -334,7 +338,7 @@ sub add_signal_presence {
my @ids_from_sigp = parse_gff3($gfffh);

open my $outfh, '>',
File::Spec->catfile( $outdir, 'contigs_prot_reg_new.fas' )
File::Spec->catfile( $tmpdir, 'contigs_prot_reg_new.fas' )
or croak "Cannot open outfile: $OS_ERROR";
_addsigpres( $outfh, $protfile, @ids_from_sigp );
close $outfh or croak "Failed to close file: $OS_ERROR";
Expand Down Expand Up @@ -460,7 +464,7 @@ sub split_by_fam {
Bio::SeqIO->new(
-format => 'fasta',
-file => '>'
. File::Spec->catfile( $outdir, 'famaln',
. File::Spec->catfile( $tmpdir, 'famaln',
$regid . '_fam.fas' )
)->write_seq($panobj);
}
Expand All @@ -477,7 +481,7 @@ sub write_output {
or croak "Print to $outfh failed: $OS_ERROR";

my $seqfile = Bio::DB::Fasta->new(
File::Spec->catfile( $outdir, 'contigs_prot_reg_new.fas' ) );
File::Spec->catfile( $tmpdir, 'contigs_prot_reg_new.fas' ) );
my $fam;
foreach my $k2 ( keys %{$hashref2} ) {

Expand Down Expand Up @@ -683,7 +687,7 @@ sub run_translation {

# Translating
$seqin = Bio::SeqIO->new(
-file => File::Spec->catfile( $outdir, $outfh ),
-file => File::Spec->catfile( $tmpdir, $outfh ),
-format => 'fasta'
);

Expand All @@ -700,7 +704,7 @@ sub run_translation {
else {
msg('Translating transcripts');
translate( $seqin,
File::Spec->catfile( $outdir, 'contigs_prot.fas' ) );
File::Spec->catfile( $tmpdir, 'contigs_prot.fas' ) );
msg('Done translating');
}
}
Expand All @@ -718,7 +722,7 @@ sub run_translation {
else {
msg('Translating transcripts');
translate( $seqin,
File::Spec->catfile( $outdir, 'contigs_prot.fas' ) );
File::Spec->catfile( $tmpdir, 'contigs_prot.fas' ) );
msg('Done translating');
}
}
Expand Down Expand Up @@ -853,6 +857,7 @@ __END__
=for stopwords Koua ekenda.hmm itap transcriptome tsc nucleotides NGS ouput contigs
=for stopwords kallisto signalp blastp blastx hmmcompete Anicet Ebou MERCHANTABILITY
=for Koualab
=head1 NAME
Expand Down Expand Up @@ -943,7 +948,7 @@ I<itap> needs kallisto, signalp, and hmmcompete to properly work.
=head1 LICENSE AND COPYRIGHT
Copyright 2019-2020 by Anicet Ebou <anicet.ebou@gmail.com>.
Copyright 2019-2020 Koualab.
This program is free software: you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
Expand Down Expand Up @@ -982,6 +987,6 @@ None known to the best of my knowledge. Please report them to the author.
=head1 AUTHOR
Anicet Ebou <anicet.ebou@gmail.com>
Dominique Koua <koua.dominique@gmail.com>
Dominique Koua <dominique.koua@inphb.ci>
=cut

0 comments on commit 987d600

Please sign in to comment.