Merge branch 'master' of github.com:bioperl/bioperl-live

bioperl · Oct 24, 2011 · f3c6e04 · f3c6e04
2 parents f20b8fd + 416a408
commit f3c6e04
Show file tree

Hide file tree

Showing 7 changed files with 81 additions and 36 deletions.
diff --git a/Bio/FeatureIO/gff.pm b/Bio/FeatureIO/gff.pm
@@ -208,50 +208,49 @@ sub next_feature_group {
   $self->{group_not_done} = 1;
 
   while ($self->{group_not_done} && ($feat = $self->next_feature()) && defined($feat)) {
-	# we start by collecting all features in the group and
-	# memorizing those which have an ID attribute
+    # we start by collecting all features in the group and
+    # memorizing those which have an ID attribute
     my $anno_ID = $feat->get_Annotations('ID');
-	if(ref($anno_ID)) {
+    if(ref($anno_ID)) {
       my $attr_ID = $anno_ID->value;
       $self->throw("Oops! ID $attr_ID exists more than once in your file!")
-		if (exists($seen_ids{$attr_ID}));
+        if (exists($seen_ids{$attr_ID}));
       $seen_ids{$attr_ID} = $feat;
-	}
-	push(@all_feats, $feat);
+    }
+    push(@all_feats, $feat);
   }
 
   # assemble the top-level features
   foreach $feat (@all_feats) {
-	my @parents = $feat->get_Annotations('Parent');
-	if (@parents) {
+    my @parents = $feat->get_Annotations('Parent');
+    if (@parents) {
       foreach my $parent (@parents) {
-		my $parent_id = $parent->value;
-		$self->throw("Parent with ID $parent_id not found!") unless (exists($seen_ids{$parent_id}));
-		$seen_ids{$parent_id}->add_SeqFeature($feat);
-      }
-	} else {
-	    push(@toplevel_feats, $feat);
+        my $parent_id = $parent->value;
+        $self->throw("Parent with ID $parent_id not found!") unless (exists($seen_ids{$parent_id}));
+        $seen_ids{$parent_id}->add_SeqFeature($feat);
       }
+    } else {
+      push(@toplevel_feats, $feat);
+    }
   }
 
   return @toplevel_feats;
 }
 
 =head2 next_seq()
 
-access the FASTA section (if any) at the end of the GFF stream.  note that this method
-will return undef if not all features in the stream have been handled
+ Usage   : $featureio->next_seq( );
+ Function: access the FASTA section (if any) at the end of the GFF stream. Note
+           that this method will return undef before all the features in the GFF
+           stream have been handled.
+ Returns : a Bio::SeqI object or undef
+ Args    : none
 
 =cut
 
 sub next_seq() {
   my $self = shift;
-  return unless $self->fasta_mode();
-
-  #first time next_seq has been called.  initialize Bio::SeqIO instance
-  if(!$self->seqio){
-    $self->seqio( Bio::SeqIO->new(-format => 'fasta', -fh => $self->_fh()) );
-  }
+  return undef unless $self->fasta_mode();
   return $self->seqio->next_seq();
 }
 
@@ -295,7 +294,8 @@ sub write_feature {
 =head2 fasta_mode()
 
  Usage   : $obj->fasta_mode($newval)
- Function: 
+ Function: Returns whether or not we're currently in a FASTA section of the GFF
+           stream
  Example : 
  Returns : value of fasta_mode (a scalar)
  Args    : on set, new value (a scalar or undef, optional)
@@ -311,7 +311,7 @@ sub fasta_mode {
   $self->{'fasta_mode'} = $val if defined($val);
 
   if ($val && $val == 1) {
-  #  seek $self->_fh(), -1, 1; #rewind 1 byte to get the previous line's \n
+    #seek $self->_fh(), -1, 1; #rewind 1 byte to get the previous line's \n
     $self->_pushback("\n");
   }
 
@@ -321,19 +321,28 @@ sub fasta_mode {
 =head2 seqio()
 
  Usage   : $obj->seqio($newval)
- Function: holds a Bio::SeqIO instance for handling the GFF3 ##FASTA section.
- Returns : value of seqio (a scalar)
+ Function: get/set a Bio::SeqIO instance to handle the GFF3 ##FASTA section.
+ Returns : a Bio::SeqIO object or undef
  Args    : on set, new value (a scalar or undef, optional)
 
-
 =cut
 
 sub seqio {
   my($self,$val) = @_;
-  $self->{'seqio'} = $val if defined($val);
+ if (defined $val) {
+    $self->{'seqio'} = $val;
+  } else {
+    # Cannot get seqio before we've reached the ##FASTA section
+    return undef unless $self->fasta_mode();
+    if (not defined $self->{'seqio'}) {
+      # Initialize Bio::SeqIO instance
+      $self->{'seqio'} = Bio::SeqIO->new(-format => 'fasta', -fh => $self->_fh());
+    }
+  }
   return $self->{'seqio'};
 }
 
+
 =head2 sequence_region()
 
  Usage   :

diff --git a/Bio/Seq/SimulatedRead.pm b/Bio/Seq/SimulatedRead.pm
@@ -248,7 +248,7 @@ sub _create_desc {
    }
    # Description of the original sequence
    my $ref_desc = $self->reference->desc;
-   if (defined $self->reference->desc) {
+   if ( (defined $self->reference->desc) && ($self->reference->desc !~ m/^\s*$/) ) {
       $ref_desc =~ s/"/\\"/g; # escape double-quotes to \"
       $desc_str .= 'description="'.$ref_desc.'" ';
    }

diff --git a/Bio/SeqFeature/Annotated.pm b/Bio/SeqFeature/Annotated.pm
@@ -719,7 +719,7 @@ sub attach_seq {
    my ($self, $seq) = @_;
 
    if ( ! ($seq && ref($seq) && $seq->isa("Bio::PrimarySeqI")) ) {
-       $self->throw("Must attach Bio::PrimarySeqI objects to SeqFeatures");
+       $self->throw("Must attach Bio::PrimarySeqI objects to SeqFeatures but got '".ref($seq)."'");
    }
 
    $self->{'seq'} = $seq;

diff --git a/Bio/SeqFeature/Generic.pm b/Bio/SeqFeature/Generic.pm
@@ -586,7 +586,7 @@ sub attach_seq {
    my ($self, $seq) = @_;
 
    if ( ! ($seq && ref($seq) && $seq->isa("Bio::PrimarySeqI")) ) {
-       $self->throw("Must attach Bio::PrimarySeqI objects to SeqFeatures");
+       $self->throw("Must attach Bio::PrimarySeqI objects to SeqFeatures but got '".ref($seq)."'");
    }
 
    $self->{'_gsf_seq'} = $seq;

diff --git a/Bio/SimpleAlign.pm b/Bio/SimpleAlign.pm
@@ -1064,6 +1064,29 @@ sub select_noncont {
 	return $aln;
 }
 
+=head2 select_noncont_by_name
+
+ Title     : select_noncont_by_name
+ Usage     : my $aln2 = $aln->select_noncont_by_name('A123', 'B456');
+ Function  : Creates a new alignment from a subset of sequences which are
+             selected by name (sequence ID).
+ Returns   : a Bio::SimpleAlign object
+ Args      : array of names (i.e., identifiers) for the sequences.
+
+=cut
+
+sub select_noncont_by_name {
+    my ($self, @names) = @_;
+
+    my $aln = $self->new;
+    foreach my $name (@names) {
+        $aln->add_seq($self->get_seq_by_id($name));
+    }
+    $aln->id($self->id);
+
+    return $aln;
+}
+
 =head2 slice
 
  Title     : slice

diff --git a/t/Align/SimpleAlign.t b/t/Align/SimpleAlign.t
@@ -7,7 +7,7 @@ BEGIN {
     use lib '.';
     use Bio::Root::Test;
 
-    test_begin( -tests => 201 );
+    test_begin( -tests => 198 );
 
     use_ok('Bio::SimpleAlign');
     use_ok('Bio::AlignIO');
@@ -38,7 +38,7 @@ my $aln2 = $aln->select( 1, 3 );
 isa_ok( $aln2, 'Bio::Align::AlignI' );
 is( $aln2->num_sequences, 3, 'num_sequences' );
 
-# test select non continuous-sorted by default
+# test select non contiguous-sorted by default
 $aln2 = $aln->select_noncont( 6, 7, 8, 9, 10, 1, 2, 3, 4, 5 );
 is( $aln2->num_sequences, 10, 'num_sequences' );
 is(
@@ -52,7 +52,7 @@ is(
     'select_noncont'
 );
 
-# test select non continuous-nosort option
+# test select non contiguous-nosort option
 $aln2 = $aln->select_noncont( 'nosort', 6, 7, 8, 9, 10, 1, 2, 3, 4, 5 );
 is( $aln2->num_sequences, 10, 'num_sequences' );
 is(
@@ -66,6 +66,15 @@ is(
     'select_noncont'
 );
 
+# test select non contiguous by name
+my $aln3 = $aln->select_noncont_by_name('1433_LYCES','BMH1_YEAST','143T_HUMAN');
+is( $aln3->num_sequences, 3, 'select_noncont_by_name' );
+my @seqs3 = $aln3->each_seq();
+is $seqs3[0]->id, '1433_LYCES', 'select_noncont_by_name';
+is $seqs3[1]->id, 'BMH1_YEAST', 'select_noncont_by_name';
+is $seqs3[2]->id, '143T_HUMAN', 'select_noncont_by_name';
+
+
 @seqs = $aln->each_seq();
 is scalar @seqs, 16, 'each_seq';
 is $seqs[0]->get_nse, '1433_LYCES/9-246', 'get_nse';

diff --git a/t/SeqFeature/FeatureIO.t b/t/SeqFeature/FeatureIO.t
@@ -7,8 +7,8 @@ BEGIN {
     use lib '.';
     use Bio::Root::Test;
 
-    test_begin(-tests => 47,
-	       -requires_module => 'Graph');
+    test_begin(-tests => 50,
+               -requires_module => 'Graph');
 
     use_ok('Bio::FeatureIO');
 }
@@ -35,6 +35,7 @@ while($f = $io->next_feature()){
 }
 is($fcount, 0);
 
+
 #then try to read sequences again.  should get seqs now
 while($s = $io->next_seq()){
     $scount++;
@@ -84,6 +85,7 @@ $scount = 0;
 ok( $io = Bio::FeatureIO->new( -file => test_input_file('hybrid1.gff3') ) );
 
 #try to read sequences first.  should be undef
+is $io->seqio, undef;
 while($s = $io->next_seq()){
     $scount++;
 }
@@ -96,8 +98,10 @@ while($f = $io->next_feature()){
 is($fcount , 6);
 
 #then try to read sequences again.
+isa_ok $io->seqio, 'Bio::SeqIO';
 while($s = $io->next_seq()){
     $scount++;
+    isa_ok $s, 'Bio::Seq';
     TODO: {
 	local $TODO = 'How did this ever work?!?';
 	if ($scount == 1) {