Merge branch 'master' into branch-1-6

bioperl · Jul 29, 2010 · 40eb18f · 40eb18f
2 parents 0c61488 + f63c4db
commit 40eb18f
Show file tree

Hide file tree

Showing 4 changed files with 108 additions and 9 deletions.
diff --git a/Bio/DB/SeqFeature/Store/DBI/Pg.pm b/Bio/DB/SeqFeature/Store/DBI/Pg.pm
@@ -327,9 +327,9 @@ END
    typeid            int not null,
    seqid             int not null,
    bin               int not null,
-   cum_count         int not null,
-   CREATE UNIQUE INDEX interval_stats_id ON interval_stats(typeid,seqid,bin)
-)
+   cum_count         int not null
+ );
+ CREATE UNIQUE INDEX interval_stats_id ON interval_stats(typeid,seqid,bin);
 END
 	 };
 }
@@ -934,9 +934,28 @@ sub _add_interval_stats_table {
     my $self = shift;
     my $tables          = $self->table_definitions;
     my $interval_stats  = $self->_interval_stats_table;
-    eval {
-	$self->dbh->do("CREATE TABLE $interval_stats $tables->{interval_stats}");
+    ##check to see if it exists yet; if it does, just return because
+    ##there is a drop from in the next step
+    my $dbh = $self->dbh;
+    my @table_exists = $dbh->selectrow_array("SELECT * FROM pg_tables WHERE tablename
+ = '$interval_stats' AND schemaname = '".$self->namespace."'");
+    if (!scalar(@table_exists)) {
+        my $query = "CREATE TABLE $interval_stats $tables->{interval_stats}";
+        $dbh->do($query) or $self->throw($dbh->errstr);
     }
+}    
+
+sub _fetch_indexed_features_sql {
+    my $self     = shift;
+    my $features = $self->_feature_table;
+    return <<END;
+SELECT typeid,seqid,start-1,"end"
+  FROM $features as f
+ WHERE f.indexed=1
+  ORDER BY typeid,seqid,start
+END
 }
 
+
+
 1;
diff --git a/Bio/DB/SeqFeature/Store/DBI/mysql.pm b/Bio/DB/SeqFeature/Store/DBI/mysql.pm
@@ -1756,7 +1756,31 @@ sub _rebuild_obj {
 
         $attribs{$attribute} = $attribute_value;
     }
-
+
+    # if we weren't called with all the params, pull those out of the database too
+    if ( not ( grep { defined($_) } ( $typeid, $db_seqid, $start, $end, $strand ))) {
+      my $sql = qq{ SELECT start,end,tag,strand,seqname
+                    FROM feature,feature_location,typelist,locationlist
+                    WHERE feature.id=feature_location.id AND feature.typeid=typelist.id
+                    AND seqid=locationlist.id AND feature.id = ? };
+      my $sth = $self->_prepare($sql) or $self->throw($self->dbh->errstr);
+      $sth->execute($id);
+      my ($feature_start, $feature_end, $feature_type, $feature_strand,$feature_seqname);
+      $sth->bind_columns(\($feature_start, $feature_end, $feature_type, $feature_strand, $feature_seqname));
+      while ($sth->fetch()) {
+        # there should be only one row returned, but we call like this to
+        # ensure we get all rows
+      }
+      $start  ||= $feature_start;
+      $end    ||= $feature_end;
+      $strand ||= $feature_strand;
+      $seqid  ||= $feature_seqname;
+
+      my( $feature_typename , $feature_typesource ) = split /:/ , $feature_type;
+      $type   ||= $feature_typename;
+      $source ||= $feature_typesource;
+    }
+
     my $obj = Bio::SeqFeature::Lite->new(-primary_id => $id,
                                          $type ? (-type => $type) : (),
                                          $source ? (-source => $source) : (),
@@ -1765,7 +1789,7 @@ sub _rebuild_obj {
                                          defined $end ? (-end => $end) : (),
                                          defined $strand ? (-strand => $strand) : (),
                                          keys %attribs ? (-attributes => \%attribs) : ());
-    
+
     return $obj;
 }
 

diff --git a/scripts/Bio-SeqFeature-Store/bp_seqfeature_gff3.PLS b/scripts/Bio-SeqFeature-Store/bp_seqfeature_gff3.PLS
@@ -4,6 +4,7 @@
 
 use strict;
 
+use Carp;
 use Getopt::Long;
 use File::Spec;
 use Bio::DB::SeqFeature::Store;
@@ -55,7 +56,20 @@ $SIG{TERM} = $SIG{INT} = sub {   undef $store; die "Aborted..."; };
 
 my $seq_stream = $store->get_seq_stream(@ARGV)  or die "failed to get_seq_stream(@ARGV)"; 
 while (my $seq = $seq_stream->next_seq) {
-  print $seq->gff3_string(@gff3opt) . "\n";
+  ### 20100725 // genehack
+  # Try to call a gff3_string() method, but fall back to gff_string() if $seq
+  # doesn't support that. Note that gff_string() is required per
+  # Bio::SeqFeatureI, while gff3_string() is not. Currently, only
+  # Bio::SeqFeature::Lite implements gff3_string().
+  if ( $seq->can( 'gff3_string' )) {
+    print $seq->gff3_string(@gff3opt) . "\n";
+  }
+  elsif ( $seq->can( 'gff_string' )) {
+    print $seq->gff_string . "\n";
+  }
+  else {
+    confess "sequence object $seq does not support gff3_string() or gff_string() methods!"
+  }
 }
 
 exit 0;
diff --git a/t/SeqFeature/SeqFeature.t b/t/SeqFeature/SeqFeature.t
@@ -7,7 +7,7 @@ BEGIN {
     use lib '.';
     use Bio::Root::Test;
 
-    test_begin(-tests => 222);
+    test_begin(-tests => 245);
 
 	use_ok('Bio::Seq');
 	use_ok('Bio::SeqIO');
@@ -377,3 +377,45 @@ lives_ok {
   is_deeply( [$sf->get_tagset_values('notag') ], [], 'get_tagset_values no tag values found');
 } 'get_tagset_values lives with no tag';
 throws_ok { $sf->get_tag_values('notag') } qr/tag value that does not exist/, 'get_tag_values throws with no tag';
+
+# circular sequence SeqFeature tests
+$seqin = Bio::SeqIO->new(-format => 'genbank',
+                         -file   => test_input_file('PX1CG.gb'));
+
+$seq = $seqin->next_seq;
+ok($seq->is_circular, 'Phi-X174 genome is circular');
+
+# retrieving the spliced sequence from any split location requires
+# spliced_seq()
+
+my %sf_data = (
+    #       start
+    'A'  => [3981, 136, 1, 1542, 'join(3981..5386,1..136)', 'ATGGTTCGTT'],
+    'A*' => [4497, 136, 1, 1026, 'join(4497..5386,1..136)', 'ATGAAATCGC'],
+    'B'  => [5075, 136, 1, 363,  'join(5075..5386,1..51)',  'ATGGAACAAC'],
+);
+
+my @split_sfs = grep {
+    $_->location->isa('Bio::Location::SplitLocationI')
+    } $seq->get_SeqFeatures();
+
+is(@split_sfs, 3, 'only 3 split locations');
+
+for my $sf (@split_sfs) {
+    isa_ok($sf->location, 'Bio::Location::SplitLocationI');
+    my ($tag) = $sf->get_tag_values('product');
+    my ($start, $end, $strand, $length, $ftstring, $first_ten) = @{$sf_data{$tag}};
+
+    # these pass
+    is($sf->location->to_FTstring, $ftstring, 'feature string');
+    is($sf->spliced_seq->subseq(1,10), $first_ten, 'first ten nucleotides');
+        is($sf->strand, $strand, 'strand');
+
+    TODO: {
+        local $TODO = "Need to define how to deal with start, end length for circular sequences";
+        is($sf->start, $start, 'start');
+        is($sf->end, $end, 'end');
+        is($sf->length, $length, 'expected length');
+    }
+}
+