added non-overlapping feature to split_seq

git-svn-id: http://biopieces.googlecode.com/svn/trunk@525 74ccb610-7750-0410-82ae-013aeee3265d

diff --git a/bp_bin/split_seq b/bp_bin/split_seq
index 7c4c5980a6e4162bc8e85262a629f7ee63a9cd16..274e13306376b98be145540dc9b582519d66e746 100755 (executable)
--- a/bp_bin/split_seq
+++ b/bp_bin/split_seq
@@ -33,23 +33,27 @@ use Maasha::Biopieces;
 # >>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>><<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<
 
 
-my ( $options, $in, $out, $record, $new_record, $i, $subseq, %lookup );
+my ( $options, $in, $out, $record, $new_record, $i, $inc, $subseq, %lookup );
 
 $options = Maasha::Biopieces::parse_options(
     [
-        { long => 'word_size', short => 'w', type => 'uint', mandatory => 'no', default => 7,     allowed => undef, disallowed => 0 },
-        { long => 'uniq',      short => 'u', type => 'flag', mandatory => 'no', default => undef, allowed => undef, disallowed => undef },
+        { long => 'word_size',       short => 'w', type => 'uint', mandatory => 'no', default => 7,     allowed => undef, disallowed => 0 },
+        { long => 'non_overlapping', short => 'n', type => 'flag', mandatory => 'no', default => undef, allowed => undef, disallowed => undef },
+        { long => 'uniq',            short => 'u', type => 'flag', mandatory => 'no', default => undef, allowed => undef, disallowed => undef },
     ]   
 );
 
 $in  = Maasha::Biopieces::read_stream( $options->{ "stream_in" } );
 $out = Maasha::Biopieces::write_stream( $options->{ "stream_out" } );
 
+$inc = 1;
+$inc = $options->{ "word_size" } if $options->{ "non_overlapping" };
+
 while ( $record = Maasha::Biopieces::get_record( $in ) ) 
 {
     if ( $record->{ "SEQ_NAME" } and $record->{ "SEQ" } )
     {
-        for ( $i = 0; $i < length( $record->{ "SEQ" } ) - $options->{ "word_size" } + 1; $i++ )
+        for ( $i = 0; $i < length( $record->{ "SEQ" } ) - $options->{ "word_size" } + 1; $i += $inc )
         {
             $subseq = substr $record->{ "SEQ" }, $i, $options->{ "word_size" };
 
@@ -57,6 +61,7 @@ while ( $record = Maasha::Biopieces::get_record( $in ) )
             {
                 $new_record->{ "SEQ_NAME" } = $record->{ "SEQ_NAME" } . "[" . ( $i + 1 ) . "-" . ( $i + $options->{ "word_size" } ) . "]";
                 $new_record->{ "SEQ" }      = $subseq;
+                $new_record->{ "SEQ_LEN" }  = $options->{ "word_size" };
 
                 Maasha::Biopieces::put_record( $new_record, $out );
 
@@ -66,6 +71,7 @@ while ( $record = Maasha::Biopieces::get_record( $in ) )
             {
                 $new_record->{ "SEQ_NAME" } = $record->{ "SEQ_NAME" } . "[" . ( $i + 1 ) . "-" . ( $i + $options->{ "word_size" } ) . "]";
                 $new_record->{ "SEQ" }      = $subseq;
+                $new_record->{ "SEQ_LEN" }  = $options->{ "word_size" };
 
                 Maasha::Biopieces::put_record( $new_record, $out );
             }