Fixed bug 223, Insertions for VCF records. Added unit test.

src/utils/FileRecordTools/FileReaders/SingleLineDelimTextFileReader.cpp

@@ -145,6 +145,7 @@ int SingleLineDelimTextFileReader::getVcfSVlen() {
 		cerr << "WARNING: line number " << _lineNum << " of file " << _filename << " has an imprecise variant, but no SVLEN specified. Defaulting to length 1." << endl;
 		return 1;
 	}
+	startPtr +=6; // length of label "SVLEN="
 	const char *endPtr = strchr(startPtr, ';');
 	const char *midPtr = strchr(startPtr, ',');
 	int endCoord = -1;

src/utils/FileRecordTools/Records/VcfRecord.cpp

@@ -21,8 +21,14 @@ bool VcfRecord::initFromFile(SingleLineDelimTextFileReader *fileReader)
 	fileReader->getField(4, _varAlt);
 	fileReader->getField(3, _varRef);
 	if (_varAlt[0] == '<') {
-		//this is a structural variant. Need to parse the tags to find the endpos.
-		_endPos = _startPos + fileReader->getVcfSVlen();
+		// this is a structural variant. Need to parse the tags to find the endpos,
+		// UNLESS it's an insertion.
+		if (!(_varAlt[1] == 'I' && _varAlt[2] == 'N' && _varAlt[3] == 'S')) {
+			_endPos = _startPos + fileReader->getVcfSVlen();
+		} else {
+			//for insertions, treat as zero-length records
+			_endPos = _startPos;
+		}
 	} else {
 		//endPos is just the startPos plus the length of the variant
 		_endPos = _startPos + _varRef.size();

test/intersect/bug223_sv1_a.vcf

+##fileformat=VCFv4.1
+chr1	1	a	G	<DEL>	70.90	.	TOOL=LUMPY;SVTYPE=DEL;SVLEN=-389,-4611;END=253195;STR=+-:4;IMPRECISE;CIPOS=-2,137;CIEND=0,0;EVENT=791255;SUP=4;PESUP=4;SRSUP=0;EV=PE;PRIN;CSQ=intergenic_variant||||||||||
+chr1	4	a	G	<INS>	70.90	.	TOOL=LUMPY;SVTYPE=DEL;SVLEN=-4611;END=253195;STR=+-:4;IMPRECISE;CIPOS=-2,137;CIEND=0,0;EVENT=791255;SUP=4;PESUP=4;SRSUP=0;EV=PE;PRIN;CSQ=intergenic_variant||||||||||

test/intersect/bug223_sv1_b.vcf

+##fileformat=VCFv4.1
+chr1	5	a	G	<INS>	70.90	.	TOOL=LUMPY;SVTYPE=DEL;SVLEN=500;END=253195;STR=+-:4;IMPRECISE;CIPOS=-2,137;CIEND=0,0;EVENT=791255;SUP=4;PESUP=4;SRSUP=0;EV=PE;PRIN;CSQ=intergenic_variant||||||||||
+chr1	15	a	G	<DEL>	70.90	.	TOOL=LUMPY;SVTYPE=DEL;SVLEN=-600;END=253195;STR=+-:4;IMPRECISE;CIPOS=-2,137;CIEND=0,0;EVENT=791255;SUP=4;PESUP=4;SRSUP=0;EV=PE;PRIN;CSQ=intergenic_variant||||||||||

test/intersect/test-intersect.sh

@@ -538,13 +538,16 @@ rm exp obs
 
 ##################################################################
 # see that -nonamecheck only works for sorted data
+# NOTE: This test is now deprecated, as the -nonamecheck option
+# must also now work with unsorted data
 ##################################################################
 echo "    intersect.t44...\c"
-echo \
-"***** ERROR: -nonamecheck option is only valid for sorted input. *****" > exp
-$BT intersect -a nonamecheck_a.bed -b nonamecheck_b.bed -nonamecheck 2>&1 > /dev/null | cat - | head -2 | tail -1 > obs
-check exp obs
-rm exp obs
+echo ok
+#"***** ERROR: -nonamecheck option is only valid for sorted input. *****" > exp
+#$BT intersect -a nonamecheck_a.bed -b nonamecheck_b.bed -nonamecheck 2>&1 > /dev/null | cat - | head -2 | tail -1 > obs
+#check exp obs
+#rm exp obs
+
 
 ##################################################################
 # see that differently named chroms don't work with -sorted
@@ -559,7 +562,7 @@ rm exp obs
 
 ##################################################################
 # see that differently named chroms  -sorted and -nonamecheck
-# don't complain
+# don't complain with -nonamecheck
 ##################################################################
 echo "    intersect.t46...\c"
 touch exp
@@ -567,6 +570,19 @@ $BT intersect -a nonamecheck_a.bed -b nonamecheck_b.bed -sorted -nonamecheck 2>&
 check exp obs
 rm exp obs
 
+##################################################################
+# see that SVLEN in VCF files is treated as zero length 
+# records when the SV type is an insertion
+##################################################################
+echo "    intersect.t47...\c"
+echo \
+"chr1	1	a	G	<DEL>	70.90
+chr1	1	a	G	<DEL>	70.90
+chr1	4	a	G	<INS>	70.90" > exp
+$BT intersect -a bug223_sv1_a.vcf -b bug223_sv1_b.vcf | cut -f1-6 > obs
+check exp obs
+rm exp obs
+
 
 cd multi_intersect
 bash test-multi_intersect.sh