Merge pull request #64 from nkindlon/master

Fixed merge bugs w/ sort order and chrom change, added unit tests

src/utils/Contexts/ContextBase.cpp

@@ -101,6 +101,7 @@ bool ContextBase::determineOutputType() {
 	//Otherwise, if the input is BAM, then the output is BAM
 	if (getFile(0)->getFileType() == FileRecordTypeChecker::BAM_FILE_TYPE) {
 		setOutputFileType(FileRecordTypeChecker::BAM_FILE_TYPE);
+		return true;
 	}
 
 	//Okay, it's bed.
@@ -264,11 +265,19 @@ int ContextBase::getBamHeaderAndRefIdx() {
 		//already found which BAM file to use for the header
 		return _bamHeaderAndRefIdx;
 	}
-	if (_files[_queryFileIdx]->getFileType() == FileRecordTypeChecker::BAM_FILE_TYPE) {
-		_bamHeaderAndRefIdx = _queryFileIdx;
-	} else {
-		_bamHeaderAndRefIdx = _databaseFileIdx;
+	if (hasIntersectMethods()) {
+		if (_files[_queryFileIdx]->getFileType() == FileRecordTypeChecker::BAM_FILE_TYPE) {
+			_bamHeaderAndRefIdx = _queryFileIdx;
+		} else {
+			_bamHeaderAndRefIdx = _databaseFileIdx;
+		}
+		return _bamHeaderAndRefIdx;
 	}
+	if (_files[0]->getFileType() == FileRecordTypeChecker::BAM_FILE_TYPE) {
+		_bamHeaderAndRefIdx = 0;
+		return _bamHeaderAndRefIdx;
+	}
+
 	return _bamHeaderAndRefIdx;
 }
 

src/utils/Contexts/ContextMerge.cpp

@@ -10,8 +10,10 @@
 
 ContextMerge::ContextMerge()
 {
+	setSortedInput(true);
 	setUseMergedIntervals(true);
 	setColumnOpsMethods(true);
+	setExplicitBedOutput(true);
 
 	//merge has no default columnOps the way map does, so we'll need to clear those.
 	_keyListOps->setColumns("");

src/utils/FileRecordTools/FileRecordMergeMgr.cpp

@@ -95,16 +95,14 @@ Record *FileRecordMergeMgr::getNextRecord(RecordKeyList *recList)
 		//check that we are still on the same chromosome.
 		const QuickString &newChrom = nextRecord->getChrName();
 		if (newChrom != currChrom) { //hit a different chromosome.
-			if (_foundChroms.find(newChrom) == _foundChroms.end() || takenFromStorage) {
-				//haven't seen this chromosome before, sort order is already enforced in the base class method.
-				if (!mustDelete) {
-					addToStorage(nextRecord);
-				} else {
-					deleteRecord(nextRecord);
-				}
-				nextRecord = NULL;
-				break;
+			//haven't seen this chromosome before, sort order is already enforced in the base class method.
+			if (!mustDelete) {
+				addToStorage(nextRecord);
+			} else {
+				deleteRecord(nextRecord);
 			}
+			nextRecord = NULL;
+			break;
 		}
 
 		//check whether it's in range

src/utils/RecordOutputMgr/RecordOutputMgr.cpp

@@ -76,18 +76,18 @@ void RecordOutputMgr::init(ContextBase *context) {
 }
 
 bool RecordOutputMgr::printKeyAndTerminate(RecordKeyList &keyList) {
+	if (_context->getProgram() == ContextBase::MERGE) {
+		//when printing merged records, we want to force the printing into
+		//bed3 format, which is surprisingly difficult to do. Had to use the following:
+		const Bed3Interval *bed3 = static_cast<const Bed3Interval *>(keyList.getKey());
+		bed3->Bed3Interval::print(_outBuf);
+		return false;
+	}
 	printBamType bamCode = printBamRecord(keyList);
 	if (bamCode == BAM_AS_BAM) {
 		return true;
 	} else if (bamCode == NOT_BAM) {
-		if (_context->getProgram() == ContextBase::MERGE) {
-			//when printing merged records, we want to force the printing into
-			//bed3 format, which is surprisingly difficult to do. Had to use the following:
-			const Bed3Interval *bed3 = static_cast<const Bed3Interval *>(keyList.getKey());
-			bed3->Bed3Interval::print(_outBuf);
-		} else {
-			keyList.getKey()->print(_outBuf);
-		}
+		keyList.getKey()->print(_outBuf);
 		return false;
 	}
 	//otherwise, it was BAM_AS_BED, and the key was printed.
@@ -221,17 +221,6 @@ void RecordOutputMgr::printRecord(RecordKeyList &keyList, RecordKeyList *blockLi
 		printKeyAndTerminate(keyList);
 		_currBamBlockList = NULL;
 		return;
-	} else if (_context->getProgram() == ContextBase::MERGE) {
-		if (!printKeyAndTerminate(keyList)) {
-			if (_context->getDesiredStrand() != FileRecordMergeMgr::ANY_STRAND) {
-				//add the sign of the record
-				tab();
-				_outBuf.append(keyList.getKey()->getStrand());
-			}
-			if (!_afterVal.empty()) tab();
-		}
-		_currBamBlockList = NULL;
-		return;
 	}
 }
 

test/merge/b.bed

+chr1	9	30	2
+chr1	10	20	1
+chr1	11	25	3
+chr1	12	30	5
+chr1	100	110	4
+chr2	11	20	6
+

test/merge/test-merge.sh

@@ -239,4 +239,40 @@ check exp obs
 rm obs exp
 
 
+###########################################################
+#  Test that sort order is enforced
+###########################################################
+echo "    merge.t19...\c"
+echo \
+"Error: Sorted input specified, but the file unsorted.bed has the following out of order record
+chr1	9	30	2" > exp
+$BT merge -i unsorted.bed 2>&1 > /dev/null | head -2  >obs
+check exp obs
+rm obs exp
+
+###########################################################
+#  Test that chrom change is handled correctly
+###########################################################
+echo "    merge.t20...\c"
+echo \
+"chr1	9	30
+chr1	100	110
+chr2	11	20" > exp
+$BT merge -i b.bed > obs
+check exp obs
+rm exp obs
+
+###########################################################
+#  Test that a merged BAM file only gives BED3 output
+###########################################################
+echo "    merge.t21...\c"
+echo \
+"chr1	10	20
+chr1	30	100
+chr2	10	20
+chr2	30	40
+chr2	42	100" > exp
+$BT merge -i a.full.bed > obs
+check exp obs
+rm exp obs
 

test/merge/unsorted.bed

+chr1	10	20	1
+chr1	9	30	2
+chr1	11	25	3
+chr1	100	110	4
+chr1	12	30	5
+chr2	11	20	6
+