changed uclust engine from CD-hit to uclust

git-svn-id: http://biopieces.googlecode.com/svn/trunk@837 74ccb610-7750-0410-82ae-013aeee3265d

diff --git a/bp_bin/cluster_seq b/bp_bin/cluster_seq
index 26037e1adaad98e1c0c8c7f7d7954bf18aa481b7..04fc8bb045ce40bee0589e752dc6d51a245c02c0 100755 (executable)
--- a/bp_bin/cluster_seq
+++ b/bp_bin/cluster_seq
@@ -39,7 +39,7 @@ use Maasha::Filesys;
 # >>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>><<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<
 
 
-my ( $options, $in, $out, $tmp_dir, $tmp_fh1, $tmp_fh2, $fh, $record, $entry, @args, $arg_str, $clusters );
+my ( $options, $in, $out, $tmp_dir, $tmp_fh1, $tmp_fh2, $fh, $record, $entry, @args1, @args2, $arg_str1, $arg_str2, $clusters );
 
 $tmp_dir  = Maasha::Biopieces::get_tmpdir();
 
@@ -72,16 +72,24 @@ while ( $record = Maasha::Biopieces::get_record( $in ) )
 close $tmp_fh1;
 close $tmp_fh2;
 
-push @args, "--ucluster $tmp_dir/cluster.fasta";
-push @args, "--id $options->{ 'identity' }";
-push @args, "--tmpdir $options->{ 'tmp_dir' }";
-push @args, "--output $tmp_dir/cluster.uc";
-push @args, "--quiet"          if not $options->{ 'verbose' };
-push @args, "> /dev/null 2>&1" if not $options->{ 'verbose' };
+push @args1, "--sort $tmp_dir/cluster.fasta";
+push @args1, "--output $tmp_dir/cluster.fasta.sort";
+push @args1, "--tmpdir $options->{ 'tmp_dir' }";
+push @args1, "--quiet"          if not $options->{ 'verbose' };
+push @args1, "> /dev/null 2>&1" if not $options->{ 'verbose' };
 
-$arg_str = join " ", @args;
+push @args2, "--input $tmp_dir/cluster.fasta.sort";
+push @args2, "--id $options->{ 'identity' }";
+push @args2, "--tmpdir $options->{ 'tmp_dir' }";
+push @args2, "--uc $tmp_dir/cluster.uc";
+push @args2, "--quiet"          if not $options->{ 'verbose' };
+push @args2, "> /dev/null 2>&1" if not $options->{ 'verbose' };
 
-Maasha::Common::run( "uclust", $arg_str );
+$arg_str1 = join " ", @args1;
+$arg_str2 = join " ", @args2;
+
+Maasha::Common::run( "uclust", $arg_str1 );
+Maasha::Common::run( "uclust", $arg_str2 );
 
 $clusters = parse_clusters( "$tmp_dir/cluster.uc" );
 
@@ -121,9 +129,11 @@ sub parse_clusters
 
     while ( $line = <$fh> )
     {
+        next if $line =~ /^#/;
+
         chomp $line;
 
-        ( $cluster, undef, undef, undef, $seq_name ) = split "\t", $line;
+        ( undef, $cluster, undef, undef, undef, undef, undef, undef, $seq_name ) = split "\t", $line;
 
         $clusters{ $seq_name } = $cluster;
     }