X-Git-Url: https://git.donarmstrong.com/?a=blobdiff_plain;f=bp_bin%2Fbowtie_seq;h=2b9254d882c8ac728f6c59b2ec2c2dc9c9bc49c5;hb=65dca98102521647b2ccb2168b095bfc808a9c40;hp=c69d29c9f8d71560c94c911351320aeb90cc97ec;hpb=7473792b77113ead9bf77a7b4d184464f3b7f41c;p=biopieces.git

diff --git a/bp_bin/bowtie_seq b/bp_bin/bowtie_seq
index c69d29c..2b9254d 100755
--- a/bp_bin/bowtie_seq
+++ b/bp_bin/bowtie_seq
@@ -42,16 +42,26 @@ my ( $options, $in, $out, $index, $tmp_dir, $tmp_in, $tmp_out, $fh_in, $fh_out,
 
 $options = Maasha::Biopieces::parse_options(
     [
-        { long => 'genome',     short => 'g', type => 'genome', mandatory => 'yes', default => undef, allowed => undef,     disallowed => undef },
-        { long => 'mismatches', short => 'm', type => 'uint',   mandatory => 'no',  default => 2,     allowed => "0,1,2,3", disallowed => undef },
-        { long => 'max_hits',   short => 'h', type => 'uint',   mandatory => 'no',  default => undef, allowed => undef,     disallowed => 0     },
+        { long => 'genome',      short => 'g', type => 'genome', mandatory => 'no', default => undef, allowed => undef,     disallowed => undef },
+        { long => 'index_name',  short => 'i', type => 'string', mandatory => 'no', default => undef, allowed => undef,     disallowed => undef },
+        { long => 'mismatches',  short => 'm', type => 'uint',   mandatory => 'no', default => 0,     allowed => "0,1,2,3", disallowed => undef },
+        { long => 'max_hits',    short => 'h', type => 'uint',   mandatory => 'no', default => undef, allowed => undef,     disallowed => 0     },
+        { long => 'seed_length', short => 's', type => 'uint',   mandatory => 'no', default => 28,    allowed => undef,     disallowed => 0     },
+        { long => 'cpus',        short => 'c', type => 'uint',   mandatory => 'no', default => 1,     allowed => undef,     disallowed => 0     },
     ]   
 );
 
+Maasha::Common::error( qq(both --index_hame and --genome specified) ) if     $options->{ "genome" } and     $options->{ "index_name" };
+Maasha::Common::error( qq(no --index_name or --genome specified) )    if not $options->{ "genome" } and not $options->{ "index_name" };
+
 $in  = Maasha::Biopieces::read_stream( $options->{ "stream_in" } );
 $out = Maasha::Biopieces::write_stream( $options->{ "stream_out" } );
 
-$index = "$ENV{ 'BP_DATA' }/genomes/$options->{ 'genome' }/bowtie/$options->{ 'genome' }";
+if ( defined $options->{ 'genome' } ) {
+    $index = "$ENV{ 'BP_DATA' }/genomes/$options->{ 'genome' }/bowtie/$options->{ 'genome' }";
+} elsif (defined $options->{ 'index_name' } ) {
+    $index = $options->{ 'index_name' };
+}
 
 $tmp_dir = Maasha::Biopieces::get_tmpdir();
 $tmp_in  = "$tmp_dir/bowtie.seq";
@@ -82,7 +92,9 @@ while ( $record = Maasha::Biopieces::get_record( $in ) )
 close $fh_out;
 
 push @args, "-n $options->{ 'mismatches' }";
+push @args, "-v $options->{ 'mismatches' }";  # DANGER: using seed mismatches as alignment mismatches - may work, may not!
 push @args, "-f" if $type eq "FASTA";
+push @args, "-p $options->{ 'cpus' }";
 
 if ( defined $options->{ 'max_hits' } ) {
     push @args, "-k $options->{ 'max_hits' }";
@@ -92,9 +104,13 @@ if ( defined $options->{ 'max_hits' } ) {
 
 $arg = join " ", @args;
 
-if ( $options->{ 'verbose' } ) {
+if ( $options->{ 'verbose' } )
+{
+    print STDERR qq(Running: bowtie $arg $index $tmp_in $tmp_out\n);
     Maasha::Common::run( "bowtie", "$arg $index $tmp_in $tmp_out" );
-} else {
+}
+else
+{
     Maasha::Common::run( "bowtie", "$arg $index $tmp_in $tmp_out > /dev/null 2>&1" );
 }
 
@@ -107,7 +123,6 @@ while ( $line = <$fh_in> )
     chomp $line;
 
     @fields = split /\t/, $line;
-
     $record = bowtie2biopiece( \@fields );
 
     Maasha::Biopieces::put_record( $record, $out );
@@ -137,14 +152,14 @@ sub bowtie2biopiece
 
     my ( $record, @scores );
 
-    $record->{ 'Q_ID' }     = $entry->[ 0 ];
-    $record->{ 'STRAND' }   = $entry->[ 1 ];
-    $record->{ 'S_ID' }     = $entry->[ 2 ];
-    $record->{ 'S_BEG' }    = $entry->[ 3 ];
-    $record->{ 'SEQ' }      = $entry->[ 4 ];
-    $record->{ 'SCORES' }   = $entry->[ 5 ];
-    $record->{ 'MISMATCH' } = $entry->[ 6 ];
-
+    $record->{ 'Q_ID' }       = $entry->[ 0 ];
+    $record->{ 'STRAND' }     = $entry->[ 1 ];
+    $record->{ 'S_ID' }       = $entry->[ 2 ];
+    $record->{ 'S_BEG' }      = $entry->[ 3 ];
+    $record->{ 'SEQ' }        = $entry->[ 4 ];
+    $record->{ 'SCORES' }     = $entry->[ 5 ];
+    $record->{ 'SCORE' }      = $entry->[ 6 ] + 1;
+    $record->{ 'DESCRIPTOR' } = $entry->[ 7 ];
     $record->{ 'SEQ_LEN' }    = length $entry->[ 4 ];
     $record->{ 'S_END' }      = $record->{ 'S_BEG' } + $record->{ 'SEQ_LEN' } - 1;
     $record->{ 'SCORES' }     =~ s/(.)/ord( $1 ) - 33 . ";"/ge; # http://maq.sourceforge.net/fastq.shtml