By @rpinho

Refereed correctness bugs in open-source research software, with patches and links to upstream PRs

bf54d4b / cpan-GO-TermFinder/patches/all-six.patch

Raw
diff --git a/lib/GO/AnnotationProvider/AnnotationParser.pm b/lib/GO/AnnotationProvider/AnnotationParser.pm
index d196016..1201240 100755
--- a/lib/GO/AnnotationProvider/AnnotationParser.pm
+++ b/lib/GO/AnnotationProvider/AnnotationParser.pm
@@ -580,7 +580,7 @@ sub __init{
 
 	@line = split("\t", $_, -1);
 
-	next if $line[$kNotColumn] eq 'NOT'; # skip annotations NOT to a GOID
+	next if $line[$kNotColumn] =~ /(?:^|\|)NOT(?:\||$)/; # skip annotations NOT to a GOID
 
 	($databaseId, $goid, $aspect) = @line[$kDatabaseIdColumn, $kGoidColumn, $kAspectColumn];
 	($standardName, $aliases)     = @line[$kStandardNameColumn, $kAliasesColumn];
diff --git a/lib/GO/OntologyProvider/OboParser.pm b/lib/GO/OntologyProvider/OboParser.pm
index b13a61a..9f79a2a 100644
--- a/lib/GO/OntologyProvider/OboParser.pm
+++ b/lib/GO/OntologyProvider/OboParser.pm
@@ -273,18 +273,27 @@ sub __init { # okay
     my @entryLine;
 
     my $isValidEntry = 0;
+    my $isObsolete = 0;
 
     my $namespace;
+    my $skipStanza = 0;
 
     while (<$ontologyFh>){
 
 	chomp;
 
-	# finish parsing the obo file of we reach the typedef line.
-
-	last if (/^\[Typedef\]/);       
+	# Finish the previous term and skip non-term stanzas, wherever they occur.
+        if (/^\[(?!Term\])[^]]+\]$/){
+            $self->__processNode(\@entryLine) if $isValidEntry;
+            @entryLine = ();
+            $isValidEntry = 0;
+            $namespace = q{};
+            $skipStanza = 1;
+            next;
+        }       
 
 	if ($_ eq '[Term]') {
+            $skipStanza = 0;
 
 	    # we reached a new term - so process the previous entry
 
@@ -299,25 +308,28 @@ sub __init { # okay
 	    @entryLine = ();
 
 	    $isValidEntry = 0;
+            $isObsolete = 0;
 	    $namespace    = '';
 
+	}elsif ($skipStanza){
+            next;
 	}elsif ($_ eq "namespace: $aspect"){
 
 	    # term is in the requested namespace
 
 	    $namespace = $aspect;
 
-	    $isValidEntry = 1;
+	    $isValidEntry = !$isObsolete;
 	
 	}elsif ($_ eq 'is_obsolete: true'){
 
 	    # we don't want obsolete nodes - DO NOT COMMENT THIS OUT -
 	    # infinite recursion will result!
 
-	    # Note, the logic here relies on the is_obsolete line coming after the
-	    # namespace line.
+	    # Keep the obsolete flag regardless of the order of tags in this stanza.
 
 	    $isValidEntry = 0;
+            $isObsolete = 1;
 		
 	}else {
 
@@ -615,6 +627,10 @@ sub __findAncestors {
 
     foreach my $ancestorGoid (@{$self->{$kParent}{$parentGoid}}) {
 
+        die "Cycle in ontology involving $ancestorGoid\n"
+            if $ancestorGoid eq $childNode->goid ||
+               grep { defined($_) && $_->goid eq $ancestorGoid } @{$pathArrayRef};
+
 	# add the ancestor node to our path to the root which is being
 	# built
 
diff --git a/lib/GO/TermFinder.pm b/lib/GO/TermFinder.pm
index 0104b3e..bb662c8 100644
--- a/lib/GO/TermFinder.pm
+++ b/lib/GO/TermFinder.pm
@@ -1771,7 +1771,9 @@ sub __samplingPopulation{
 
     if ($self->__isUsingPopulation){
 
-	@names = @{$self->__population};
+	# Resolve aliases and duplicates exactly as for the background counts.
+	my ($ids) = $self->__determineDatabaseIdsFromGenes($self->__population);
+	@names = @{$ids};
 
     }else{
 
@@ -1974,8 +1976,16 @@ sub __calculateFDR{
     # real data.  We threshold this at a maximum of 1, as we can't
     # have a FDR of greater than 100%
 
+    my $numDiscoveries = 0;
     foreach (my $i = 0; $i < @realHypotheses; $i++){
 
+        # A p-value cutoff includes every member of a tie.
+        $numDiscoveries = $i + 1 if $numDiscoveries <= $i;
+        while ($numDiscoveries < @realHypotheses &&
+               $realHypotheses[$numDiscoveries]->{PVALUE} == $realHypotheses[$i]->{PVALUE}){
+            $numDiscoveries++;
+        }
+
 	if (exists $realHypotheses[$i]->{FDR_OBSERVATIONS}){
 
 	    # the rate is the average number in the simulations that
@@ -1984,7 +1994,7 @@ sub __calculateFDR{
 
 	    $realHypotheses[$i]->{FDR_OBSERVATIONS} /= $numSimulations;
 
-	    $realHypotheses[$i]->{FDR_RATE} = $realHypotheses[$i]->{FDR_OBSERVATIONS} / ($i + 1);
+	    $realHypotheses[$i]->{FDR_RATE} = $realHypotheses[$i]->{FDR_OBSERVATIONS} / $numDiscoveries;
 
 	    if ($realHypotheses[$i]->{FDR_RATE} > 1){
 
@@ -2009,7 +2019,7 @@ sub __calculateFDR{
 	# be chosen at this point, we can calculate the expected number of
 	# false positives, as the FDR x the number of hypotheses
 
-	$realHypotheses[$i]->{EXPECTED_FALSE_POSITIVES} = $realHypotheses[$i]->{FDR_RATE} * ($i+1);	
+	$realHypotheses[$i]->{EXPECTED_FALSE_POSITIVES} = $realHypotheses[$i]->{FDR_RATE} * $numDiscoveries;	
 
     }