adapted to work with tandem repeats

more mutilation of James' beautiful adaptor ...

modules/Bio/EnsEMBL/DBSQL/RepeatFeatureAdaptor.pm

@@ -12,7 +12,7 @@ use vars qw(@ISA);
 
 sub fetch_by_RawContig {
     my( $self, $contig ) = @_;
-    
+
     my @repeats = $self->fetch_by_contig_id($contig->dbID);
     foreach my $r (@repeats) {
         $r->attach_seq($contig);
@@ -22,7 +22,7 @@ sub fetch_by_RawContig {
 
 sub fetch_by_contig_id {
     my( $self, $contig_id ) = @_;
-    
+
     return $self->_generic_fetch(
         qq{ contig_id = $contig_id }
         );
@@ -30,7 +30,7 @@ sub fetch_by_contig_id {
 
 sub fetch_by_dbID {
     my( $self, $db_id ) = @_;
-    
+
     my ($rf) = $self->_generic_fetch(
         qq{ repeat_feature_id = $db_id }
         );
@@ -39,7 +39,7 @@ sub fetch_by_dbID {
 
 sub _generic_fetch {
     my( $self, $where_clause ) = @_;
-    
+
     my( $repeat_feature_id,
         $contig_id,
         $contig_start,
@@ -50,7 +50,7 @@ sub _generic_fetch {
         $repeat_end,
         $analysis_id,
         );
-    
+
     my $sth = $self->prepare(qq{
         SELECT repeat_feature_id
           , contig_id
@@ -63,7 +63,7 @@ sub _generic_fetch {
           , analysis_id
         FROM repeat_feature f
         WHERE }. $where_clause);
-    
+
     $sth->execute;
     $sth->bind_columns(
         \$repeat_feature_id,
@@ -76,7 +76,7 @@ sub _generic_fetch {
         \$repeat_end,
         \$analysis_id,
         );
-    
+
     my $rca = $self->db->get_RepeatConsensusAdaptor;
     my $aa  = $self->db->get_AnalysisAdaptor;
 
@@ -85,18 +85,18 @@ sub _generic_fetch {
         # new in RepeatFeature takes no arguments
         my $r = Bio::EnsEMBL::RepeatFeature->new;
         $r->dbID($repeat_feature_id);
-        
+
         # So RepeatFeature can get its repeat
         $r->repeat_consensus_adaptor($rca);
         $r->repeat_id($repeat_id);
-        
+
         $r->contig_id( $contig_id     );
         $r->start    ( $contig_start  );
         $r->end      ( $contig_end    );
         $r->strand   ( $contig_strand );
         $r->hstart   ( $repeat_start  );
         $r->hend     ( $repeat_end    );
-        
+
         my( $ana_obj );
         unless ($ana_obj = $analysis_cache{$analysis_id}) {
             $ana_obj = $aa->fetch_by_dbID($analysis_id)
@@ -104,7 +104,7 @@ sub _generic_fetch {
             $analysis_cache{$analysis_id} = $ana_obj;
         }
         $r->analysis($ana_obj);
-        
+
         push(@repeats, $r);
     }
     return( @repeats );
@@ -115,7 +115,7 @@ sub store {
 
     my $rca = $self->db->get_RepeatConsensusAdaptor;
     my ($cons, $db_id);
-    
+
     $self->throw("Can't store repeats without a contig_id (got '$contig_id')")
         unless $contig_id =~ /^\d+$/;
 
@@ -134,31 +134,47 @@ sub store {
         });
     foreach my $rf (@repeats) {
 
-        unless ($rf->repeat_id) {
+	# must have a consensus attached
+
+        $self->throw("Must have a RepeatConsensus attached")
+	 unless defined ($cons = $rf->repeat_consensus);
 
-	    # need to get the consensus seq object for this repeat
+	# for tandem repeats - simply store consensus and repeat
+	# one pair per hit. don't need to check consensi stored
+	# already. consensus has name and class set to 'trf'
+
+	if ($cons->name eq 'trf') {
+	    $rca->store($cons);
+	    $rf->repeat_id($cons->dbID);
 
-            $self->throw("Must have a RepeatConsensus attached")
-	     unless defined ($cons = $rf->repeat_consensus);
+	} else {
 
-            unless ($cons->dbID) {
-	        my @match = ($rca->fetch_by_name($cons->name));
+	# for other repeats - need to see if a consensus is stored already
 
-		if (@match > 1) {
-		    $self->warn(@match . " consensi for " . $cons->name . "\n");
-		}
-		# FIXME - need to take some action here if we don't
-		# match a consensus seq already stored
-		elsif (@match == 0) {
-		    $self->warn("Can't find " . $cons->name . "\n");
-		    $cons->repeat_consensus("N");
-		    $rca->store($cons);
-		}
-	        $db_id = ($rca->fetch_by_name($cons->name))[0]->dbID;
+            unless ($rf->repeat_id) {
 
-	        $cons->dbID($db_id);
+	        # need to get the consensus seq object for this repeat
+
+                unless ($cons->dbID) {
+	            my @match = ($rca->fetch_by_name($cons->name));
+
+		    if (@match > 1) {
+		        $self->warn(@match . " consensi for " . $cons->name . "\n");
+		    } elsif (@match == 0) {
+		    # if we don't match a consensus already stored
+		    # create a fake one
+		    # set consensus to 'N' as null seq not allowed
+		    # FIXME: not happy with this, but ho hum ...
+		        $self->warn("Can't find " . $cons->name . "\n");
+		        $cons->repeat_consensus("N");
+		        $rca->store($cons);
+		    }
+	            $db_id = ($rca->fetch_by_name($cons->name))[0]->dbID;
+
+	            $cons->dbID($db_id);
+	        }
+	        $rf->repeat_id($cons->dbID);
 	    }
-	    $rf->repeat_id($cons->dbID);
 	}
 
         $sth->execute(