added 5prime adaptor feature to remove_adaptor

git-svn-id: http://biopieces.googlecode.com/svn/trunk@259 74ccb610-7750-0410-82ae-013aeee3265d

diff --git a/code_perl/Maasha/Biopieces.pm b/code_perl/Maasha/Biopieces.pm
index ab7add7e4adb71c953f954918a0c2b9f72345e20..5de534e7abe5af94c9aa9198e8908acccac186f9 100644 (file)
--- a/code_perl/Maasha/Biopieces.pm
+++ b/code_perl/Maasha/Biopieces.pm
@@ -747,7 +747,7 @@ sub get_options
         @options = qw(
             adaptor|a=s
             mismatches|m=s
-            no_remove|n
+            remove|r=s
             offset|o=s
         );
     }
@@ -1048,6 +1048,10 @@ sub get_options
         {
             Maasha::Common::error( qq(Argument to --$opt must be octal or decimal - not "$options{ $opt }") );
         }
+        elsif ( $opt eq "remove" and $script eq "remove_adaptor" and $options{ $opt } !~ /before|after|skip/ )
+        {
+            Maasha::Common::error( qq(Argument to --$opt must be before, after, or skip - not "$options{ $opt }") );
+        }
     }
 
     Maasha::Common::error( qq(no --database specified) )                if $script eq "create_blast_db"     and not $options{ "database" };
@@ -4696,6 +4700,8 @@ sub script_remove_adaptor
 
     my ( $record, $adaptor, $seq, $adaptor_len, $seq_len, $offset, $max_match, $max_mismatch, $pos );
 
+    $options->{ "remove" } ||= "after";
+
     $max_mismatch = $options->{ "mismatches" } || 0;
     $offset       = $options->{ "offset" };
 
@@ -4719,10 +4725,18 @@ sub script_remove_adaptor
 
             $record->{ "ADAPTOR_POS" } = $pos;
 
-            if ( $pos >= 0 and not $options->{ "no_remove" } )
+            if ( $pos >= 0 and $options->{ "remove" } ne "skip" )
             {
-                $record->{ "SEQ" }     = substr $record->{ "SEQ" }, 0, $pos;
-                $record->{ "SEQ_LEN" } = $pos;
+                if ( $options->{ "remove" } eq "after" )
+                {
+                    $record->{ "SEQ" }     = substr $record->{ "SEQ" }, 0, $pos;
+                    $record->{ "SEQ_LEN" } = $pos;
+                }
+                else
+                {
+                    $record->{ "SEQ" }     = substr $record->{ "SEQ" }, $pos + $adaptor_len;
+                    $record->{ "SEQ_LEN" } = length $record->{ "SEQ" };
+                }
             }
 
             put_record( $record, $out );
@@ -5583,7 +5597,7 @@ sub script_plot_histogram
 
     while ( $record = get_record( $in ) ) 
     {
-        $data_hash{ $record->{ $options->{ "key" } } }++ if $record->{ $options->{ "key" } };
+        $data_hash{ $record->{ $options->{ "key" } } }++ if defined $record->{ $options->{ "key" } };
 
         put_record( $record, $out ) if not $options->{ "no_stream" };
     }
@@ -5623,7 +5637,7 @@ sub script_plot_lendist
 
     while ( $record = get_record( $in ) ) 
     {
-        $data_hash{ $record->{ $options->{ "key" } } }++ if $record->{ $options->{ "key" } };
+        $data_hash{ $record->{ $options->{ "key" } } }++ if defined $record->{ $options->{ "key" } };
 
         put_record( $record, $out ) if not $options->{ "no_stream" };
     }