App-Dazz
view release on metacpan or search on metacpan
lib/App/Dazz/Command/overlap.pm view on Meta::CPAN
push @infiles, Path::Tiny::path($infile)->absolute->stringify;
}
if ( lc $opt->{outfile} ne "stdout" ) {
$opt->{outfile} = Path::Tiny::path( $opt->{outfile} )->absolute->stringify;
}
# record cwd, we'll return there
my $cwd = Path::Tiny->cwd;
my $tempdir;
if ( $opt->{tmp} ) {
$tempdir = Path::Tiny->tempdir(
TEMPLATE => "dazz_ovlp_XXXXXXXX",
DIR => $opt->{tmp},
);
}
else {
$tempdir = Path::Tiny->tempdir("dazz_ovlp_XXXXXXXX");
}
chdir $tempdir;
my $basename = $tempdir->basename();
$basename =~ s/\W+/_/g;
{ # Preprocess reads to format them for dazzler
my $cmd = "cat";
$cmd .= sprintf " %s", $_ for @infiles;
$cmd .= " | dazz dazzname stdin -o stdout";
$cmd .= " | faops filter -l 0 stdin renamed.fasta";
App::Dazz::Common::exec_cmd( $cmd, { verbose => $opt->{verbose}, } );
if ( !$tempdir->child("stdout.replace.tsv")->is_file ) {
Carp::croak "Failed: create stdout.replace.tsv\n";
}
}
{ # Make the dazzler DB, each block is of size 50 MB
my $cmd;
$cmd .= "fasta2DB $basename renamed.fasta";
$cmd .= " && DBdust $basename";
$cmd .= " && DBsplit -s50 $basename";
App::Dazz::Common::exec_cmd( $cmd, { verbose => $opt->{verbose}, } );
if ( !$tempdir->child("$basename.db")->is_file ) {
Carp::croak "Failed: fasta2DB\n";
}
}
{ # Run daligner
my $block_number;
for my $line ( $tempdir->child("$basename.db")->lines ) {
if ( $line =~ /^blocks\s+=\s+(\d+)/ ) {
$block_number = $1;
last;
}
}
my $cmd;
$cmd
.= "HPC.daligner $basename -M16 -T$opt->{parallel} -e$opt->{idt} -l$opt->{len} -s$opt->{len} -mdust";
$cmd .= " | sed 's/ -vS / -S /'"; # don't show LAcheck prompts
$cmd .= " | bash";
App::Dazz::Common::exec_cmd( $cmd, { verbose => $opt->{verbose}, } );
if ( defined $block_number and $block_number > 1 ) {
$cmd = "LAcat $basename.\@.las > $basename.las";
App::Dazz::Common::exec_cmd( $cmd, { verbose => $opt->{verbose}, } );
}
if ( !$tempdir->child("$basename.las")->is_file ) {
Carp::croak "Failed: daligner\n";
}
}
{ # outputs
my $cmd = "LAshow -o $basename.db $basename.las > show.txt";
if ( $opt->{all} ) {
$cmd = "LAshow $basename.db $basename.las > show.txt";
}
App::Dazz::Common::exec_cmd( $cmd, { verbose => $opt->{verbose}, } );
if ( !$tempdir->child("show.txt")->is_file ) {
Carp::croak "Failed: LAshow\n";
}
$cmd = "dazz show2ovlp renamed.fasta show.txt";
if ( !$opt->{serial} ) {
$cmd .= " -r stdout.replace.tsv";
}
$cmd .= " -o $opt->{outfile}";
App::Dazz::Common::exec_cmd( $cmd, { verbose => $opt->{verbose}, } );
}
chdir $cwd;
}
1;
( run in 1.255 second using v1.01-cache-2.11-cpan-6aa56a78535 )