Ignore target range for hit culling in frameshift mode.

bbuchfink · Nov 21, 2022 · 1938c34 · 1938c34
1 parent cfc1524
commit 1938c34
Show file tree

Hide file tree

Showing 3 changed files with 17 additions and 1 deletion.
diff --git a/src/align/legacy/query_mapper.cpp b/src/align/legacy/query_mapper.cpp
@@ -310,7 +310,7 @@ void Target::inner_culling()
 		filter_evalue = DBL_MAX;
 	}
 	for (list<Hsp>::iterator i = hsps.begin(); i != hsps.end();) {
-		if (i->is_enveloped_by(hsps.begin(), i, 0.5))
+		if (i->query_range_enveloped_by(hsps.begin(), i, 0.5))
 			i = hsps.erase(i);
 		else
 			++i;

diff --git a/src/basic/hssp.cpp b/src/basic/hssp.cpp
@@ -233,6 +233,20 @@ bool Hsp::is_enveloped_by(std::list<Hsp>::const_iterator begin, std::list<Hsp>::
 	return false;
 }
 
+bool Hsp::query_range_enveloped_by(const Hsp& hsp, double p) const
+{
+	return query_source_range.overlap_factor(hsp.query_source_range) >= p;
+}
+
+bool Hsp::query_range_enveloped_by(std::list<Hsp>::const_iterator begin, std::list<Hsp>::const_iterator end, double p) const
+{
+	for (list<Hsp>::const_iterator i = begin; i != end; ++i)
+		if (query_range_enveloped_by(*i, p))
+			return true;
+	return false;
+}
+
+
 void Hsp::push_match(Letter q, Letter s, bool positive)
 {
 	if (q == s) {

diff --git a/src/basic/match.h b/src/basic/match.h
@@ -217,6 +217,8 @@ struct Hsp
 
 	bool is_enveloped_by(const Hsp &hsp, double p) const;
 	bool is_enveloped_by(std::list<Hsp>::const_iterator begin, std::list<Hsp>::const_iterator end, double p) const;
+	bool query_range_enveloped_by(const Hsp& hsp, double p) const;
+	bool query_range_enveloped_by(std::list<Hsp>::const_iterator begin, std::list<Hsp>::const_iterator end, double p) const;
 	bool is_weakly_enveloped_by(std::list<Hsp>::const_iterator begin, std::list<Hsp>::const_iterator end, int cutoff) const;
 	void push_back(const DiagonalSegment &d, const TranslatedSequence &query, const Sequence& subject, bool reversed);
 	void push_match(Letter q, Letter s, bool positive);