From 8fe692c5aebc051c41879fb431710fc4a5e8559d Mon Sep 17 00:00:00 2001
From: nkindlon <nek3d@virginia.edu>
Date: Wed, 9 Oct 2013 14:53:09 -0400
Subject: [PATCH] Moved short circuit for BAM unmapped reads to
Record::sameChromIntersect for when query record isn't BAM but DB record is.
---
src/utils/Contexts/Context.cpp | 7 ++-
src/utils/Contexts/Context.h | 4 ++
src/utils/FileRecordTools/FileRecordMgr.cpp | 2 +-
src/utils/FileRecordTools/RecordOutputMgr.cpp | 4 +-
.../FileRecordTools/Records/BamRecord.cpp | 44 +++++++++++++------
src/utils/FileRecordTools/Records/BamRecord.h | 3 --
src/utils/FileRecordTools/Records/Record.cpp | 5 +++
7 files changed, 50 insertions(+), 19 deletions(-)
diff --git a/src/utils/Contexts/Context.cpp b/src/utils/Contexts/Context.cpp
index ffc22b2d..c201fbe5 100644
--- a/src/utils/Contexts/Context.cpp
+++ b/src/utils/Contexts/Context.cpp
@@ -39,6 +39,7 @@ Context::Context()
_databaseFileIdx(-1),
_bamHeaderAndRefIdx(-1),
_maxNumDatabaseFields(0),
+ _useFullBamTags(false),
_reportCount(false),
_maxDistance(0),
_reportNames(false),
@@ -234,6 +235,10 @@ bool Context::parseCmdArgs(int argc, char **argv, int skipFirstArgs) {
setUncompressedBam(true);
markUsed(i - skipFirstArgs);
}
+ else if (strcmp(argv[i], "-fbam") == 0) {
+ setUseFullBamTags(true);
+ markUsed(i - skipFirstArgs);
+ }
else if(strcmp(argv[i], "-sorted") == 0) {
setSortedInput(true);
markUsed(i - skipFirstArgs);
@@ -327,7 +332,7 @@ bool Context::cmdArgsValid()
int Context::getBamHeaderAndRefIdx() {
if (_bamHeaderAndRefIdx != -1) {
- //already found which BAM file to usefor the header
+ //already found which BAM file to use for the header
return _bamHeaderAndRefIdx;
}
if (_inputFiles[_queryFileIdx]._fileType == FileRecordTypeChecker::BAM_FILE_TYPE) {
diff --git a/src/utils/Contexts/Context.h b/src/utils/Contexts/Context.h
index 5638e087..53000b1c 100644
--- a/src/utils/Contexts/Context.h
+++ b/src/utils/Contexts/Context.h
@@ -162,6 +162,9 @@ public:
bool getPrintable() const { return _printable; }
void setPrintable(bool val) { _printable = val; }
+ bool getUseFullBamTags() const { return _useFullBamTags; }
+ void setUseFullBamTags(bool val) { _useFullBamTags = val; }
+
//
// MERGE METHODS
//
@@ -234,6 +237,7 @@ protected:
int _databaseFileIdx;
int _bamHeaderAndRefIdx;
int _maxNumDatabaseFields;
+ bool _useFullBamTags;
bool _reportCount;
int _maxDistance;
diff --git a/src/utils/FileRecordTools/FileRecordMgr.cpp b/src/utils/FileRecordTools/FileRecordMgr.cpp
index 9bc93bdf..ab3b516b 100644
--- a/src/utils/FileRecordTools/FileRecordMgr.cpp
+++ b/src/utils/FileRecordTools/FileRecordMgr.cpp
@@ -227,7 +227,7 @@ void FileRecordMgr::allocateFileReader()
case FileRecordTypeChecker::BAM_FILE_TYPE:
_fileReader = new BamFileReader();
- (static_cast<BamFileReader *>(_fileReader))->setUseTags(_useFullBamTags);
+ (static_cast<BamFileReader *>(_fileReader))->setUseTags(_context->getUseFullBamTags());
break;
default:
break;
diff --git a/src/utils/FileRecordTools/RecordOutputMgr.cpp b/src/utils/FileRecordTools/RecordOutputMgr.cpp
index 4f10ff3b..b6b8fd29 100644
--- a/src/utils/FileRecordTools/RecordOutputMgr.cpp
+++ b/src/utils/FileRecordTools/RecordOutputMgr.cpp
@@ -114,7 +114,9 @@ void RecordOutputMgr::printRecord(RecordKeyList &keyList, RecordKeyList *blockLi
if (needsFlush()) {
flush();
}
-
+// if (keyList.getKey()->getChrName() == "chr1" && keyList.getKey()->getStartPos() == 11996) {
+// printf("Break point here.\n");
+// }
//The first time we print a record is when we print any header, because the header
//hasn't been read from the query file until after the first record has also been read.
if (_context->getPrintHeader()) {
diff --git a/src/utils/FileRecordTools/Records/BamRecord.cpp b/src/utils/FileRecordTools/Records/BamRecord.cpp
index 0ea849a3..fda5a28b 100644
--- a/src/utils/FileRecordTools/Records/BamRecord.cpp
+++ b/src/utils/FileRecordTools/Records/BamRecord.cpp
@@ -52,9 +52,37 @@ void BamRecord::clear()
{
Bed6Interval::clear();
_bamChromId = -1;
- //For now, we're going to not clear the BamAlignment object, as all of its
- //fields will be reset next time it is used anyway. If testing shows this to be a
- //problem, we'll revisit.
+
+
+ //Clear the BamAlignment object. Sadly, it does not have a clear() method,
+ //so we have to do each member manually.
+ _bamAlignment.Name.clear();
+ _bamAlignment.Length = 0;
+ _bamAlignment.QueryBases.clear();
+ _bamAlignment.AlignedBases.clear();
+ _bamAlignment.Qualities.clear();
+ _bamAlignment.TagData.clear();
+ _bamAlignment.RefID = -1;
+ _bamAlignment.Position = -1;
+ _bamAlignment.Bin = 0;
+ _bamAlignment.MapQuality = 0;
+ _bamAlignment.AlignmentFlag = 0;
+ _bamAlignment.CigarData.clear();
+ _bamAlignment.MateRefID = -1;
+ _bamAlignment.MatePosition = -1;
+ _bamAlignment.InsertSize = -1;
+ _bamAlignment.Filename.clear();
+
+ _bamAlignment.SupportData.AllCharData.clear();
+ _bamAlignment.SupportData.BlockLength = 0;
+ _bamAlignment.SupportData.NumCigarOperations = 0;
+ _bamAlignment.SupportData.QueryNameLength = 0;
+ _bamAlignment.SupportData.QuerySequenceLength = 0;
+ _bamAlignment.SupportData.HasCoreOnly = false;
+
+ _bamAlignment.ErrorString.clear();
+
+
}
void BamRecord::print(QuickString &outBuf, RecordKeyList *keyList) const
@@ -127,13 +155,3 @@ void BamRecord::printUnmapped(QuickString &outBuf) const {
outBuf.append(_score);
outBuf.append("\t.\t-1\t-1\t-1\t0,0,0\t0\t.\t."); // dot for strand, -1 for blockStarts and blockEnd
}
-
-bool BamRecord::sameChromIntersects(const Record *record,
- bool wantSameStrand, bool wantDiffStrand, float overlapFraction, bool reciprocal) const
-{
- // Special: For BAM records that are unmapped, intersect should automatically return false
- if (_isUnmapped || record->isUnmapped()) {
- return false;
- }
- return Record::sameChromIntersects(record, wantSameStrand, wantDiffStrand, overlapFraction, reciprocal);
-}
diff --git a/src/utils/FileRecordTools/Records/BamRecord.h b/src/utils/FileRecordTools/Records/BamRecord.h
index 3029b28e..262ef1ab 100644
--- a/src/utils/FileRecordTools/Records/BamRecord.h
+++ b/src/utils/FileRecordTools/Records/BamRecord.h
@@ -38,9 +38,6 @@ public:
const BamTools::BamAlignment &getAlignment() const { return _bamAlignment; }
int getBamChromId() const { return _bamChromId; }
- virtual bool sameChromIntersects(const Record *otherRecord,
- bool sameStrand, bool diffStrand, float overlapFraction, bool reciprocal) const;
-
protected:
BamTools::BamAlignment _bamAlignment;
int _bamChromId; //different from chromId, because BAM file may be in different order
diff --git a/src/utils/FileRecordTools/Records/Record.cpp b/src/utils/FileRecordTools/Records/Record.cpp
index 8ec1efd5..cb5e5f44 100644
--- a/src/utils/FileRecordTools/Records/Record.cpp
+++ b/src/utils/FileRecordTools/Records/Record.cpp
@@ -135,6 +135,11 @@ bool Record::intersects(const Record *record,
bool Record::sameChromIntersects(const Record *record,
bool wantSameStrand, bool wantDiffStrand, float overlapFraction, bool reciprocal) const
{
+ // Special: For records that are unmapped, intersect should automatically return false
+ if (_isUnmapped || record->isUnmapped()) {
+ return false;
+ }
+
//If user requested hits only on same strand, or only on different strands,
//rule out different strandedness first.
//If the strand is unknown in either case, then queries regarding strandedness
--
GitLab