Updated for new schema. fetch_all_by_Slice now handles non-chromosomal slices.

modules/Bio/EnsEMBL/Lite/SNPAdaptor.pm

@@ -51,60 +51,112 @@ use vars '@ISA';
 sub fetch_all_by_Slice {
   my ($self, $slice ) = @_;
 
-  my $slice_start = $slice->chr_start();
-  my $slice_end   = $slice->chr_end();
-  
-  my %SNPS = qw( 12 dbSNP 13 WI 14 HGBASE 15 TSC-CSHL 16 ANO );
-  my $QUERY = "select internal_id, chr_start, chr_end, chr_strand, type, range_type,
-		      validated, alleles, snpclass, mapweight, ambiguity, source,
-		      id_refsnp, id_wi, id_hgbase, id_tsc, id_ano 
-                 FROM snp
-                WHERE chr_name = ? AND chr_start >= ? and chr_start <= ? AND chr_end >= ?";
-
-  my $sth = $self->prepare( $QUERY ); 
-  eval {
-  $sth->execute($slice->chr_name(), $slice_start - 500 , $slice_end, $slice_start);
-  };
-  return [] if $@;
-  my @snps = ();  
-
-  my %link_hash;
-  my $link;
+  my @snps;
+
+  my $from_strand = $slice->strand();
+
+  # wherever this slice is, it needs to be converted to
+  # a toplevel slices since all snps in the lite database 
+  # are stored on toplevel seqregions
+
+  my @projection = @{$slice->project('toplevel')};
+
+  my $link_col_idx = 12;
+  my @link_dbs = ('dbSNP', 'WI', 'HGBASE', 'TSC-CSHL', 'ANO');
+
+  $self->{'link_cache'} ||= {};
+
+  foreach my $segment (@projection) {
+    my $from_start = $segment->from_start();
+    my $from_end   = $segment->from_end();
+    my $top_slice  = $segment->to_Slice();
+    my $top_slice_start  = $top_slice->start();
+    my $top_slice_end    = $top_slice->end();
+    my $top_slice_strand = $top_slice->strand();
+
+    my $sth = $self->prepare
+      ("SELECT internal_id, chr_start, chr_end, chr_strand, type, " .
+       "       range_type, validated, alleles, snpclass, mapweight, ".
+       "       ambiguity, source, id_refsnp, id_wi, id_hgbase, id_tsc, " .
+       "       id_ano " .
+       "FROM   snp " .
+       "WHERE  chr_name = ? " .
+       "AND    chr_start >= ? " .
+       "AND    chr_start <= ? " .
+       "AND    chr_end >= ?");
+
+    $sth->execute($top_slice->seq_region_name(),
+                  $top_slice_start - 500,
+                  $top_slice_end,
+                  $top_slice_start);
+
+    while(my $arrayref = $sth->fetchrow_arrayref()) {
+      my @links = ();
+
+      # loop over the last columns of the row to retrieve a single external
+      # database link for each column
+
+      for(my $i = 0; $i < scalar(@link_dbs); $i++) {
+        my $link_id = $arrayref->[$link_col_idx + $i];
+
+        next if(!$link_id);
+
+        my $link_db = $link_dbs[$i];
+        my $link = $self->{'link_cache'}->{"$link_db:$link_id"};
+
+        if(!$link) {
+          $link = Bio::EnsEMBL::DBEntry->new_fast
+            ({'_dbname' => $link_db,
+              '_primary_id' => $link_id});
+          $self->{'link_cache'}->{"$link_db:$link_id"};
+        }
+
+        push @links, $link;
+      }
+
+      #create a snp object through a fast (hacky) constructor
+      my $status = $arrayref->[6];
+      $status =~ s/-/ /;
+      if($status && $status ne 'no info') {
+        $status = "proven $status";
+      } else {
+        $status = 'suspected';
+      }
+
+      # coordinates must be adjusted so that they are first 
+      # relative to the start of the top level slice (rather than absolute)
+      # and then adjusted so they are relative to the start of the
+      # original requested slice (w/ from_start)
+      my($start,$end,$strand);
+      if($top_slice_strand == 1) {
+        $start = $arrayref->[1] - $top_slice_start + $from_start;
+        $end   = $arrayref->[2] - $top_slice_start + $from_start;
+        $strand = $arrayref->[3];
+      } else {
+        $start = $top_slice_end - $arrayref->[2] + $from_start;
+        $end   = $top_slice_end - $arrayref->[1] + $from_start;
+        $strand = $arrayref->[3] * -1;
+      }
+
+      push @snps, Bio::EnsEMBL::SNP->new_fast
+        ({ 'dbID'       => $arrayref->[0],
+           '_gsf_start'  => $start,
+           '_gsf_end'    => $end,
+           '_snp_strand' => $strand,
+           '_gsf_score'  => 1,
+           '_type'       => $arrayref->[4],
+           '_range_type' => $arrayref->[5],
+           '_validated'  => $arrayref->[6],
+           'status'     => $status,
+           'alleles'    => $arrayref->[7],
+           '_ambiguity_code' => $arrayref->[10],
+           '_snpclass'   => $arrayref->[8],
+           '_mapweight'  => $arrayref->[9],
+           '_source' => $arrayref->[11],
+           '_source_tag' => $arrayref->[11],
+           'link'        => \@links });
 
-  while(my $arrayref = $sth->fetchrow_arrayref()) {
-    my @links = ();
-    foreach( sort keys %SNPS ) {
-       my $V = $arrayref->[ $_ ];
-       if( $V && $V ne '' ) {
-         unless($link = $link_hash{"$SNPS{$_}:$V"}) {
-           $link_hash{"$SNPS{$_}:$V"} = $link = Bio::EnsEMBL::DBEntry->new_fast( {'_dbname'     => $SNPS{$_}, '_primary_id' => $V });
-         }
-         push @links, $link;
-       }
     }
-
-    #create a snp object through a fast (hacky) constructor
-    my $STATUS = $arrayref->[6];
-    $STATUS =~s/-/ /;
-    $STATUS = ( $STATUS && $STATUS ne 'no info' ) ? "proven $STATUS" : 'suspected';
-	my $snp = Bio::EnsEMBL::SNP->new_fast(
-		  { 'dbID'       => $arrayref->[0], 
-		   '_gsf_start'  => $arrayref->[1] - $slice_start + 1,
-		    '_gsf_end'    => $arrayref->[2] - $slice_start + 1,
-		    '_snp_strand' => $arrayref->[3],
-		    '_gsf_score'  => 1,
-		    '_type'       => $arrayref->[4],
-                    '_range_type' => $arrayref->[5],
-                    '_validated'  => $arrayref->[6],
-                    'status'     => $STATUS,
-					'alleles'    => $arrayref->[7],
-                    '_ambiguity_code' => $arrayref->[10],
-                    '_snpclass'   => $arrayref->[8],
-                    '_mapweight'  => $arrayref->[9],
-		    '_source' => $arrayref->[11],
-			'_source_tag' => $arrayref->[11],
-		    'link'        => \@links });
-    push @snps, $snp;
   }
 	
   return \@snps;