From 696f8a89b29c34b783da0086c79a1ca2edbbc521 Mon Sep 17 00:00:00 2001
From: Aaron <aaronquinlan@gmail.com>
Date: Wed, 7 Dec 2011 21:32:02 -0500
Subject: [PATCH] Use new GetNextBed() interface. Redux.
---
src/annotateBed/annotateBed.cpp | 6 +-
src/bed12ToBed6/bed12ToBed6.cpp | 6 +-
src/bedToBam/bedToBam.cpp | 10 ++--
src/bedToIgv/bedToIgv.cpp | 7 +--
src/closestBed/closestBed.cpp | 6 +-
src/coverageBed/coverageBed.cpp | 8 +--
src/fastaFromBed/fastaFromBed.cpp | 6 +-
src/flankBed/flankBed.cpp | 11 +---
src/genomeCoverageBed/genomeCoverageBed.cpp | 10 ++--
src/intersectBed/intersectBed.cpp | 7 +--
src/linksBed/linksBed.cpp | 8 +--
src/multiBamCov/multiBamCov.cpp | 6 +-
src/nucBed/nucBed.cpp | 6 +-
src/shuffleBed/shuffleBed.cpp | 22 +++-----
src/slopBed/slopBed.cpp | 11 +---
src/subtractBed/subtractBed.cpp | 8 +--
src/utils/bedFile/bedFile.cpp | 59 +++++++++-----------
src/utils/bedFile/bedFile.h | 8 ++-
src/utils/chromsweep/chromsweep.cpp | 62 +++++++++------------
src/utils/chromsweep/chromsweep.h | 10 +---
src/windowBed/windowBed.cpp | 8 +--
21 files changed, 109 insertions(+), 176 deletions(-)
diff --git a/src/annotateBed/annotateBed.cpp b/src/annotateBed/annotateBed.cpp
index 0914acc9..1e5366cf 100644
--- a/src/annotateBed/annotateBed.cpp
+++ b/src/annotateBed/annotateBed.cpp
@@ -114,13 +114,11 @@ void BedAnnotate::AnnotateBed() {
for (size_t annoIndex = 0; annoIndex < _annoFiles.size(); ++annoIndex) {
// grab the current annotation file.
BedFile *anno = _annoFiles[annoIndex];
- int lineNum = 0;
BED a, nullBed;
- BedLineStatus bedStatus;
// process each entry in the current anno file
- while ((bedStatus = anno->GetNextBed(a, lineNum)) != BED_INVALID) {
- if (bedStatus == BED_VALID) {
+ while (anno->GetNextBed(a)) {
+ if (anno->_status == BED_VALID) {
_bed->countListHits(a, annoIndex, _sameStrand, _diffStrand);
a = nullBed;
}
diff --git a/src/bed12ToBed6/bed12ToBed6.cpp b/src/bed12ToBed6/bed12ToBed6.cpp
index 6726cc34..8858dbf0 100644
--- a/src/bed12ToBed6/bed12ToBed6.cpp
+++ b/src/bed12ToBed6/bed12ToBed6.cpp
@@ -136,12 +136,10 @@ void ProcessBed(istream &bedInput, BedFile *bed) {
// process each BED entry and convert to BAM
BED bedEntry, nullBed;
- int lineNum = 0;
- BedLineStatus bedStatus;
// open the BED file for reading.
bed->Open();
- while ((bedStatus = bed->GetNextBed(bedEntry, lineNum)) != BED_INVALID) {
- if (bedStatus == BED_VALID) {
+ while (bed->GetNextBed(bedEntry)) {
+ if (bed->_status == BED_VALID) {
bedVector bedBlocks; // vec to store the discrete BED "blocks" from a
splitBedIntoBlocks(bedEntry, bedBlocks);
diff --git a/src/bedToBam/bedToBam.cpp b/src/bedToBam/bedToBam.cpp
index 1c717612..3fab9af0 100644
--- a/src/bedToBam/bedToBam.cpp
+++ b/src/bedToBam/bedToBam.cpp
@@ -185,19 +185,17 @@ void ProcessBed(BedFile *bed, GenomeFile *genome, bool isBED12, int mapQual, boo
// process each BED entry and convert to BAM
BED bedEntry, nullBed;
- int lineNum = 0;
- BedLineStatus bedStatus;
// open the BED file for reading.
bed->Open();
- while ((bedStatus = bed->GetNextBed(bedEntry, lineNum)) != BED_INVALID) {
- if (bedStatus == BED_VALID) {
+ while (bed->GetNextBed(bedEntry)) {
+ if (bed->_status == BED_VALID) {
BamAlignment bamEntry;
if (bed->bedType >= 4) {
- ConvertBedToBam(bedEntry, bamEntry, chromToId, isBED12, mapQual, lineNum);
+ ConvertBedToBam(bedEntry, bamEntry, chromToId, isBED12, mapQual, bed->_lineNum);
writer->SaveAlignment(bamEntry);
}
else {
- cerr << "Error: BED entry without name found at line: " << lineNum << ". Exiting!" << endl;
+ cerr << "Error: BED entry without name found at line: " << bed->_lineNum << ". Exiting!" << endl;
exit (1);
}
bedEntry = nullBed;
diff --git a/src/bedToIgv/bedToIgv.cpp b/src/bedToIgv/bedToIgv.cpp
index 7aab2191..ec0e3f6a 100644
--- a/src/bedToIgv/bedToIgv.cpp
+++ b/src/bedToIgv/bedToIgv.cpp
@@ -224,13 +224,10 @@ void ProcessBed(istream &bedInput, BedFile *bed, string path, string sortType, s
BED bedEntry, nullBed;
- int lineNum = 0;
- BedLineStatus bedStatus;
-
bed->Open();
// process each BED entry and convert to an IGV request
- while ((bedStatus = bed->GetNextBed(bedEntry, lineNum)) != BED_INVALID) {
- if (bedStatus == BED_VALID) {
+ while (bed->GetNextBed(bedEntry)) {
+ if (bed->_status == BED_VALID) {
string filename = bedEntry.chrom + "_" + ToString(bedEntry.start) + "_" + ToString(bedEntry.end);
string locus = bedEntry.chrom + ":" + ToString(bedEntry.start - slop) + "-" + ToString(bedEntry.end + slop);
diff --git a/src/closestBed/closestBed.cpp b/src/closestBed/closestBed.cpp
index 735e2a9e..2cf6bd4f 100644
--- a/src/closestBed/closestBed.cpp
+++ b/src/closestBed/closestBed.cpp
@@ -215,10 +215,8 @@ void BedClosest::FindClosestBed() {
_bedB->loadBedFileIntoMap();
BED a, nullBed;
- int lineNum = 0; // current input line number
vector<BED> hits; // vector of potential hits
hits.reserve(100);
- BedLineStatus bedStatus;
_bedA->Open();
// report A's header first if asked.
@@ -226,8 +224,8 @@ void BedClosest::FindClosestBed() {
_bedA->PrintHeader();
}
// process each entry in A in search of the closest feature in B
- while ((bedStatus = _bedA->GetNextBed(a, lineNum)) != BED_INVALID) {
- if (bedStatus == BED_VALID) {
+ while (_bedA->GetNextBed(a)) {
+ if (_bedA->_status == BED_VALID) {
FindWindowOverlaps(a, hits);
hits.clear();
a = nullBed;
diff --git a/src/coverageBed/coverageBed.cpp b/src/coverageBed/coverageBed.cpp
index b6163516..f3e71d52 100644
--- a/src/coverageBed/coverageBed.cpp
+++ b/src/coverageBed/coverageBed.cpp
@@ -51,14 +51,11 @@ void BedCoverage::CollectCoverageBed() {
// that we can easily compare "A" to it for overlaps
_bedB->loadBedCovFileIntoMap();
- int lineNum = 0; // current input line number
BED a, nullBed;
- BedLineStatus bedStatus;
-
_bedA->Open();
// process each entry in A
- while ((bedStatus = _bedA->GetNextBed(a, lineNum)) != BED_INVALID) {
- if (bedStatus == BED_VALID) {
+ while (_bedA->GetNextBed(a)) {
+ if (_bedA->_status == BED_VALID) {
// process the BED entry as a single block
if (_obeySplits == false)
_bedB->countHits(a, _sameStrand, _diffStrand, _countsOnly);
@@ -66,7 +63,6 @@ void BedCoverage::CollectCoverageBed() {
else {
bedVector bedBlocks;
splitBedIntoBlocks(a, bedBlocks);
-
// use countSplitHits to avoid over-counting each split chunk
// as distinct read coverage.
_bedB->countSplitHits(bedBlocks, _sameStrand, _diffStrand, _countsOnly);
diff --git a/src/fastaFromBed/fastaFromBed.cpp b/src/fastaFromBed/fastaFromBed.cpp
index 41ddebd8..ce7eb203 100644
--- a/src/fastaFromBed/fastaFromBed.cpp
+++ b/src/fastaFromBed/fastaFromBed.cpp
@@ -104,13 +104,11 @@ void Bed2Fa::ExtractDNA() {
fr->open(_dbFile, memmap);
BED bed, nullBed;
- int lineNum = 0;
- BedLineStatus bedStatus;
string sequence;
_bed->Open();
- while ((bedStatus = _bed->GetNextBed(bed, lineNum)) != BED_INVALID) {
- if (bedStatus == BED_VALID) {
+ while (_bed->GetNextBed(bed)) {
+ if (_bed->_status == BED_VALID) {
// make sure we are extracting >= 1 bp
if (bed.zeroLength == false) {
size_t seqLength = fr->sequenceLength(bed.chrom);
diff --git a/src/flankBed/flankBed.cpp b/src/flankBed/flankBed.cpp
index f7e12495..14ef0ab5 100644
--- a/src/flankBed/flankBed.cpp
+++ b/src/flankBed/flankBed.cpp
@@ -41,9 +41,7 @@ BedFlank::~BedFlank(void) {
void BedFlank::FlankBed() {
- int lineNum = 0;
BED bedEntry, nullBed; // used to store the current BED line from the BED file.
- BedLineStatus bedStatus;
_bed->Open();
// report A's header first if asked.
@@ -51,10 +49,8 @@ void BedFlank::FlankBed() {
_bed->PrintHeader();
}
- bedStatus = _bed->GetNextBed(bedEntry, lineNum);
- while (bedStatus != BED_INVALID) {
- if (bedStatus == BED_VALID) {
-
+ while (_bed->GetNextBed(bedEntry)) {
+ if (_bed->_status == BED_VALID) {
int leftFlank = _leftFlank;
int rightFlank = _rightFlank;
if (_fractional == true) {
@@ -70,9 +66,8 @@ void BedFlank::FlankBed() {
{
AddStrandedFlank(bedEntry, leftFlank, rightFlank);
}
- bedEntry = nullBed;
}
- bedStatus = _bed->GetNextBed(bedEntry, lineNum);
+ bedEntry = nullBed;
}
_bed->Close();
}
diff --git a/src/genomeCoverageBed/genomeCoverageBed.cpp b/src/genomeCoverageBed/genomeCoverageBed.cpp
index a97c8964..2b960273 100644
--- a/src/genomeCoverageBed/genomeCoverageBed.cpp
+++ b/src/genomeCoverageBed/genomeCoverageBed.cpp
@@ -146,21 +146,19 @@ void BedGenomeCoverage::AddBlockedCoverage(const vector<BED> &bedBlocks) {
void BedGenomeCoverage::CoverageBed() {
BED a, nullBed;
- int lineNum = 0; // current input line number
- BedLineStatus bedStatus;
ResetChromCoverage();
_bed->Open();
- while ( (bedStatus = _bed->GetNextBed(a, lineNum)) != BED_INVALID ) {
- if (bedStatus == BED_VALID) {
+ while (_bed->GetNextBed(a)) {
+ if (_bed->_status == BED_VALID) {
if (_filterByStrand == true) {
if (a.strand.empty()) {
- cerr << "Input error: Interval is missing a strand value on line " << lineNum << "." <<endl;
+ cerr << "Input error: Interval is missing a strand value on line " << _bed->_lineNum << "." <<endl;
exit(1);
}
if ( ! (a.strand == "-" || a.strand == "+") ) {
- cerr << "Input error: Invalid strand value (" << a.strand << ") on line " << lineNum << "." << endl;
+ cerr << "Input error: Invalid strand value (" << a.strand << ") on line " << _bed->_lineNum << "." << endl;
exit(1);
}
// skip if the strand is not what the user requested.
diff --git a/src/intersectBed/intersectBed.cpp b/src/intersectBed/intersectBed.cpp
index 6a48b880..735fd8d7 100644
--- a/src/intersectBed/intersectBed.cpp
+++ b/src/intersectBed/intersectBed.cpp
@@ -199,7 +199,6 @@ void BedIntersect::IntersectBed() {
vector<BED> hits;
hits.reserve(100);
BED a, nullBed;
- BedLineStatus bedStatus;
// open the "A" file, process each BED entry and searh for overlaps.
_bedA->Open();
@@ -207,8 +206,8 @@ void BedIntersect::IntersectBed() {
if (_printHeader == true) {
_bedA->PrintHeader();
}
- while ((bedStatus = _bedA->GetNextBed(a)) != BED_INVALID) {
- if (bedStatus == BED_VALID) {
+ while (_bedA->GetNextBed(a)) {
+ if (_bedA->_status == BED_VALID) {
// treat the BED as a single "block"
if (_obeySplits == false) {
FindOverlaps(a, hits);
@@ -233,7 +232,7 @@ void BedIntersect::IntersectBed() {
}
else {
// use the chromsweep algorithm to detect overlaps on the fly.
- ChromSweep sweep = ChromSweep(_bedA, _bedB, _sameStrand, _diffStrand);
+ ChromSweep sweep = ChromSweep(_bedA, _bedB, _sameStrand, _diffStrand, _printHeader);
pair<BED, vector<BED> > hit_set;
hit_set.second.reserve(10000);
diff --git a/src/linksBed/linksBed.cpp b/src/linksBed/linksBed.cpp
index 2adb5dea..b72372b1 100644
--- a/src/linksBed/linksBed.cpp
+++ b/src/linksBed/linksBed.cpp
@@ -101,13 +101,11 @@ void BedLinks::CreateLinks() {
cout << "<table border=\"0\" align=\"justify\"" << endl;
cout << "<h3>BED Entries from: stdin </h3>" << endl;
- int lineNum = 0;
- BED bedEntry, nullBed;
- BedLineStatus bedStatus;
+ BED bedEntry, nullBed;
_bed->Open();
- while ((bedStatus = _bed->GetNextBed(bedEntry, lineNum)) != BED_INVALID) {
- if (bedStatus == BED_VALID) {
+ while (_bed->GetNextBed(bedEntry)) {
+ if (_bed->_status == BED_VALID) {
WriteURL(bedEntry, base);
bedEntry = nullBed;
}
diff --git a/src/multiBamCov/multiBamCov.cpp b/src/multiBamCov/multiBamCov.cpp
index 4c3c7396..b99774b5 100644
--- a/src/multiBamCov/multiBamCov.cpp
+++ b/src/multiBamCov/multiBamCov.cpp
@@ -58,15 +58,13 @@ void MultiCovBam::CollectCoverage()
// if index data available for all BAM files, we can use SetRegion
if ( reader.HasIndexes() ) {
BED bed, nullBed;
- int lineNum = 0;
- BedLineStatus bedStatus;
_bed->Open();
// loop through each BED entry, jump to it,
// and collect coverage from each BAM
- while ((bedStatus = _bed->GetNextBed(bed, lineNum)) != BED_INVALID)
+ while (_bed->GetNextBed(bed))
{
- if (bedStatus == BED_VALID)
+ if (_bed->_status == BED_VALID)
{
// initialize counts for each file to 0
vector<int> counts(_bam_files.size(), 0);
diff --git a/src/nucBed/nucBed.cpp b/src/nucBed/nucBed.cpp
index d5777864..f70e582b 100644
--- a/src/nucBed/nucBed.cpp
+++ b/src/nucBed/nucBed.cpp
@@ -111,13 +111,11 @@ void NucBed::ProfileDNA() {
bool headerReported = false;
BED bed, nullBed;
- int lineNum = 0;
- BedLineStatus bedStatus;
string sequence;
_bed->Open();
- while ((bedStatus = _bed->GetNextBed(bed, lineNum)) != BED_INVALID) {
- if (bedStatus == BED_VALID) {
+ while (_bed->GetNextBed(bed)) {
+ if (_bed->_status == BED_VALID) {
if (headerReported == false) {
PrintHeader();
headerReported = true;
diff --git a/src/shuffleBed/shuffleBed.cpp b/src/shuffleBed/shuffleBed.cpp
index 374459b0..48d22888 100644
--- a/src/shuffleBed/shuffleBed.cpp
+++ b/src/shuffleBed/shuffleBed.cpp
@@ -79,14 +79,10 @@ BedShuffle::~BedShuffle(void) {
void BedShuffle::Shuffle() {
-
- int lineNum = 0;
BED bedEntry, nullBed; // used to store the current BED line from the BED file.
- BedLineStatus bedStatus;
-
_bed->Open();
- while ((bedStatus = _bed->GetNextBed(bedEntry, lineNum)) != BED_INVALID) {
- if (bedStatus == BED_VALID) {
+ while (_bed->GetNextBed(bedEntry)) {
+ if (_bed->_status == BED_VALID) {
ChooseLocus(bedEntry);
_bed->reportBedNewLine(bedEntry);
bedEntry = nullBed;
@@ -99,13 +95,11 @@ void BedShuffle::Shuffle() {
void BedShuffle::ShuffleWithExclusions() {
- int lineNum = 0;
BED bedEntry, nullBed; // used to store the current BED line from the BED file.
- BedLineStatus bedStatus;
_bed->Open();
- while ((bedStatus = _bed->GetNextBed(bedEntry, lineNum)) != BED_INVALID) {
- if (bedStatus == BED_VALID) {
+ while (_bed->GetNextBed(bedEntry)) {
+ if (_bed->_status == BED_VALID) {
// keep looking as long as the chosen
// locus happens to overlap with regions
// that the user wishes to exclude.
@@ -122,7 +116,7 @@ void BedShuffle::ShuffleWithExclusions() {
if (tries > MAX_TRIES) {
- cerr << "Error, line " << lineNum << ": tried " << MAX_TRIES << " potential loci for entry, but could not avoid excluded regions. Ignoring entry and moving on." << endl;
+ cerr << "Error, line " << _bed->_lineNum << ": tried " << MAX_TRIES << " potential loci for entry, but could not avoid excluded regions. Ignoring entry and moving on." << endl;
}
else {
_bed->reportBedNewLine(bedEntry);
@@ -136,13 +130,11 @@ void BedShuffle::ShuffleWithExclusions() {
void BedShuffle::ShuffleWithInclusions() {
- int lineNum = 0;
BED bedEntry, nullBed; // used to store the current BED line from the BED file.
- BedLineStatus bedStatus;
_bed->Open();
- while ((bedStatus = _bed->GetNextBed(bedEntry, lineNum)) != BED_INVALID) {
- if (bedStatus == BED_VALID) {
+ while (_bed->GetNextBed(bedEntry)) {
+ if (_bed->_status == BED_VALID) {
// choose a new locus
ChooseLocusFromInclusionFile(bedEntry);
_bed->reportBedNewLine(bedEntry);
diff --git a/src/slopBed/slopBed.cpp b/src/slopBed/slopBed.cpp
index 68eda5a2..76257e9c 100644
--- a/src/slopBed/slopBed.cpp
+++ b/src/slopBed/slopBed.cpp
@@ -40,19 +40,15 @@ BedSlop::~BedSlop(void) {
void BedSlop::SlopBed() {
- int lineNum = 0;
BED bedEntry, nullBed; // used to store the current BED line from the BED file.
- BedLineStatus bedStatus;
_bed->Open();
// report header first if asked.
if (_printHeader == true) {
_bed->PrintHeader();
- }
-
- bedStatus = _bed->GetNextBed(bedEntry, lineNum);
- while (bedStatus != BED_INVALID) {
- if (bedStatus == BED_VALID) {
+ }
+ while (_bed->GetNextBed(bedEntry)) {
+ if (_bed->_status == BED_VALID) {
if (_fractional == false) {
AddSlop(bedEntry, (int) _leftSlop, (int) _rightSlop);
}
@@ -64,7 +60,6 @@ void BedSlop::SlopBed() {
_bed->reportBedNewLine(bedEntry);
bedEntry = nullBed;
}
- bedStatus = _bed->GetNextBed(bedEntry, lineNum);
}
_bed->Close();
}
diff --git a/src/subtractBed/subtractBed.cpp b/src/subtractBed/subtractBed.cpp
index e259787b..dded5da4 100644
--- a/src/subtractBed/subtractBed.cpp
+++ b/src/subtractBed/subtractBed.cpp
@@ -156,15 +156,13 @@ void BedSubtract::SubtractBed() {
_bedB->loadBedFileIntoMap();
BED a, nullBed;
- BedLineStatus bedStatus;
- int lineNum = 0; // current input line number
- vector<BED> hits; // vector of potential hits
+ vector<BED> hits;
// reserve some space
hits.reserve(100);
_bedA->Open();
- while ((bedStatus = _bedA->GetNextBed(a, lineNum)) != BED_INVALID) {
- if (bedStatus == BED_VALID) {
+ while (_bedA->GetNextBed(a)) {
+ if (_bedA->_status == BED_VALID) {
FindAndSubtractOverlaps(a, hits);
hits.clear();
a = nullBed;
diff --git a/src/utils/bedFile/bedFile.cpp b/src/utils/bedFile/bedFile.cpp
index 2d59aa15..9fa075f2 100644
--- a/src/utils/bedFile/bedFile.cpp
+++ b/src/utils/bedFile/bedFile.cpp
@@ -167,7 +167,7 @@ void BedFile::Seek(unsigned long offset) {
// Jump to a specific byte in the file
bool BedFile::Empty(void) {
- return _bedStream->eof();
+ return _status == BED_INVALID || _status == BED_BLANK;
}
// Close the BED file
@@ -216,7 +216,7 @@ void BedFile::PrintHeader(void) {
}
-BedLineStatus BedFile::GetNextBed(BED &bed, bool forceSorted) {
+bool BedFile::GetNextBed(BED &bed, bool forceSorted) {
// make sure there are still lines to process.
// if so, tokenize, validate and return the BED entry.
@@ -233,17 +233,16 @@ BedLineStatus BedFile::GetNextBed(BED &bed, bool forceSorted) {
_firstLine = false;
}
// load the BED struct as long as it's a valid BED entry.
- BedLineStatus status = parseLine(bed, _bedFields);
- if (!forceSorted) {
- return status;
- }
- else if (status == BED_VALID) {
+ _status = parseLine(bed, _bedFields);
+ if (_status == BED_INVALID) return false;
+
+ if (_status == BED_VALID) {
if (bed.chrom == _prev_chrom) {
if ((int) bed.start >= _prev_start) {
_prev_chrom = bed.chrom;
_prev_start = bed.start;
}
- else {
+ else if (forceSorted) {
cerr << "ERROR: input file: (" << bedFile << ") is not sorted by chrom then start" << endl;
exit(1);
}
@@ -252,15 +251,16 @@ BedLineStatus BedFile::GetNextBed(BED &bed, bool forceSorted) {
_prev_chrom = bed.chrom;
_prev_start = bed.start;
}
- return status;
- }
- else {
- return status;
+ return true;
}
+ else if (_status == BED_HEADER) {return true;}
+ // i.e, BED_BLANK
+ else {return false;}
}
// default if file is closed or EOF
- return BED_INVALID;
+ _status = BED_INVALID;
+ return false;
}
@@ -268,10 +268,9 @@ bool BedFile::GetNextMergedBed(BED &merged_bed) {
if (_bedStream->good()) {
BED bed;
- BedLineStatus bedStatus;
// force sorting; hence third param = true
- while ((bedStatus = GetNextBed(bed, true)) != BED_INVALID) {
- if (bedStatus == BED_VALID) {
+ while (GetNextBed(bed, true)) {
+ if (_status == BED_VALID) {
if (((int) bed.start - _merged_end > 0) ||
(_merged_end < 0) ||
(bed.chrom != _merged_chrom))
@@ -300,7 +299,7 @@ bool BedFile::GetNextMergedBed(BED &merged_bed) {
}
}
// handle the last merged block in the file.
- if (bedStatus == BED_INVALID)
+ if (_status == BED_INVALID)
{
merged_bed.chrom = _merged_chrom;
merged_bed.start = _merged_start;
@@ -664,12 +663,10 @@ void BedFile::setBedType (int colNums) {
void BedFile::loadBedFileIntoMap() {
BED bedEntry, nullBed;
- int lineNum = 0;
- BedLineStatus bedStatus;
Open();
- while ((bedStatus = GetNextBed(bedEntry, lineNum)) != BED_INVALID) {
- if (bedStatus == BED_VALID) {
+ while (GetNextBed(bedEntry)) {
+ if (_status == BED_VALID) {
BIN bin = getBin(bedEntry.start, bedEntry.end);
bedMap[bedEntry.chrom][bin].push_back(bedEntry);
bedEntry = nullBed;
@@ -682,12 +679,10 @@ void BedFile::loadBedFileIntoMap() {
void BedFile::loadBedCovFileIntoMap() {
BED bedEntry, nullBed;
- int lineNum = 0;
- BedLineStatus bedStatus;
Open();
- while ((bedStatus = GetNextBed(bedEntry, lineNum)) != BED_INVALID) {
- if (bedStatus == BED_VALID) {
+ while (GetNextBed(bedEntry)) {
+ if (_status == BED_VALID) {
BIN bin = getBin(bedEntry.start, bedEntry.end);
BEDCOV bedCov;
@@ -712,12 +707,10 @@ void BedFile::loadBedCovFileIntoMap() {
void BedFile::loadBedCovListFileIntoMap() {
BED bedEntry, nullBed;
- int lineNum = 0;
- BedLineStatus bedStatus;
Open();
- while ((bedStatus = GetNextBed(bedEntry, lineNum)) != BED_INVALID) {
- if (bedStatus == BED_VALID) {
+ while (GetNextBed(bedEntry)) {
+ if (_status == BED_VALID) {
BIN bin = getBin(bedEntry.start, bedEntry.end);
BEDCOVLIST bedCovList;
@@ -741,12 +734,10 @@ void BedFile::loadBedCovListFileIntoMap() {
void BedFile::loadBedFileIntoMapNoBin() {
BED bedEntry, nullBed;
- int lineNum = 0;
- BedLineStatus bedStatus;
-
+
Open();
- while ((bedStatus = this->GetNextBed(bedEntry, lineNum)) != BED_INVALID) {
- if (bedStatus == BED_VALID) {
+ while (GetNextBed(bedEntry)) {
+ if (_status == BED_VALID) {
bedMapNoBin[bedEntry.chrom].push_back(bedEntry);
bedEntry = nullBed;
}
diff --git a/src/utils/bedFile/bedFile.h b/src/utils/bedFile/bedFile.h
index 9cd78e4a..0206de8f 100644
--- a/src/utils/bedFile/bedFile.h
+++ b/src/utils/bedFile/bedFile.h
@@ -427,7 +427,7 @@ public:
void GetLine(void);
// Get the next BED entry in an opened BED file.
- BedLineStatus GetNextBed (BED &bed, bool forceSorted = false);
+ bool GetNextBed (BED &bed, bool forceSorted = false);
// Returns the next MERGED (i.e., non-overlapping) interval in an opened BED file
// NOTE: assumes input file is sorted by chrom then start
@@ -487,7 +487,9 @@ public:
masterBedCovListMap bedCovListMap;
masterBedMap bedMap;
masterBedMapNoBin bedMapNoBin;
-
+
+ BedLineStatus _status;
+ int _lineNum;
private:
// data
@@ -497,7 +499,7 @@ private:
FileType _fileType; // what is the file type? (BED? GFF? VCF?)
istream *_bedStream;
string _bedLine;
- int _lineNum;
+
string _header;
bool _firstLine;
vector<string> _bedFields;
diff --git a/src/utils/chromsweep/chromsweep.cpp b/src/utils/chromsweep/chromsweep.cpp
index 0ae73b59..70eac679 100644
--- a/src/utils/chromsweep/chromsweep.cpp
+++ b/src/utils/chromsweep/chromsweep.cpp
@@ -14,52 +14,43 @@
#include <queue>
bool after(const BED &a, const BED &b);
-void report_hits(const BED &curr_qy, const vector<BED> &hits);
-vector<BED> scan_cache(const BED &curr_qy, BedLineStatus qy_status, const vector<BED> &db_cache, vector<BED> &hits);
-
/*
// constructor using existing BedFile pointers
*/
-ChromSweep::ChromSweep(BedFile *bedA, BedFile *bedB, bool sameStrand, bool diffStrand)
-: _bedA(bedA)
-, _bedB(bedB)
+ChromSweep::ChromSweep(BedFile *query, BedFile *db, bool sameStrand, bool diffStrand, bool printHeader)
+: _query(query)
+, _db(db)
, _sameStrand(sameStrand)
, _diffStrand(diffStrand)
{
- // prime the results pump.
- _qy_lineNum = 0;
- _db_lineNum = 0;
+ _hits.reserve(100000);
+ _cache.reserve(100000);
- _hits.reserve(1000);
- _cache.reserve(1000);
+ _query->Open();
+ if (printHeader) _query->PrintHeader();
+ _db->Open();
- _bedA->Open();
- _bedB->Open();
- _qy_status = _bedA->GetNextBed(_curr_qy, _qy_lineNum);
- _db_status = _bedB->GetNextBed(_curr_db, _db_lineNum);
+ _query->GetNextBed(_curr_qy);
+ _db->GetNextBed(_curr_db);
}
/*
Constructor with filenames
*/
-ChromSweep::ChromSweep(string &bedAFile, string &bedBFile)
+ChromSweep::ChromSweep(string &queryFile, string &dbFile)
{
- // prime the results pump.
- _qy_lineNum = 0;
- _db_lineNum = 0;
-
_hits.reserve(100000);
_cache.reserve(100000);
- _bedA = new BedFile(bedAFile);
- _bedB = new BedFile(bedBFile);
+ _query = new BedFile(queryFile);
+ _db = new BedFile(dbFile);
- _bedA->Open();
- _bedB->Open();
+ _query->Open();
+ _db->Open();
- _qy_status = _bedA->GetNextBed(_curr_qy, _qy_lineNum);
- _db_status = _bedB->GetNextBed(_curr_db, _db_lineNum);
+ _query->GetNextBed(_curr_qy);
+ _db->GetNextBed(_curr_db);
}
@@ -71,7 +62,7 @@ ChromSweep::~ChromSweep(void) {
void ChromSweep::ScanCache() {
- if (_qy_status != BED_INVALID) {
+ if (_query->_status != BED_INVALID) {
vector<BED>::iterator c = _cache.begin();
while (c != _cache.end())
{
@@ -92,14 +83,13 @@ void ChromSweep::ScanCache() {
bool ChromSweep::ChromChange()
{
// the files are on the same chrom
- if ((_curr_qy.chrom == _curr_db.chrom) || (_db_status == BED_INVALID) || (_qy_status == BED_INVALID)) {
+ if (_curr_qy.chrom == _curr_db.chrom) {
return false;
}
// the query is ahead of the database. fast-forward the database to catch-up.
else if (_curr_qy.chrom > _curr_db.chrom) {
- while (!_bedB->Empty() && _curr_db.chrom < _curr_qy.chrom)
+ while (_db->GetNextBed(_curr_db, true) && _curr_db.chrom < _curr_qy.chrom)
{
- _db_status = _bedB->GetNextBed(_curr_db, _db_lineNum);
}
_cache.clear();
return false;
@@ -119,7 +109,7 @@ bool ChromSweep::ChromChange()
_results.push(make_pair(_curr_qy, _no_hits));
_cache.clear();
}
- _qy_status = _bedA->GetNextBed(_curr_qy, _qy_lineNum);
+ _query->GetNextBed(_curr_qy, true);
_curr_chrom = _curr_qy.chrom;
return true;
}
@@ -146,26 +136,28 @@ bool ChromSweep::IsValidHit(const BED &query, const BED &db) {
bool ChromSweep::Next(pair<BED, vector<BED> > &next) {
- if (!_bedA->Empty()) {
+ if (!_query->Empty()) {
// have we changed chromosomes?
if (ChromChange() == false) {
// scan the database cache for hits
ScanCache();
// advance the db until we are ahead of the query. update hits and cache as necessary
- while (!_bedB->Empty() && _curr_qy.chrom == _curr_db.chrom && !(after(_curr_db, _curr_qy)))
+ while (!_db->Empty() &&
+ _curr_qy.chrom == _curr_db.chrom &&
+ !(after(_curr_db, _curr_qy)))
{
if (IsValidHit(_curr_qy, _curr_db)) {
_hits.push_back(_curr_db);
}
_cache.push_back(_curr_db);
- _db_status = _bedB->GetNextBed(_curr_db, _db_lineNum);
+ _db->GetNextBed(_curr_db, true);
}
// add the hits for this query to the pump
_results.push(make_pair(_curr_qy, _hits));
// reset for the next query
_hits.clear();
_curr_qy = _nullBed;
- _qy_status = _bedA->GetNextBed(_curr_qy, _qy_lineNum);
+ _query->GetNextBed(_curr_qy, true);
_curr_chrom = _curr_qy.chrom;
}
}
diff --git a/src/utils/chromsweep/chromsweep.h b/src/utils/chromsweep/chromsweep.h
index af5f6798..5f83a24a 100644
--- a/src/utils/chromsweep/chromsweep.h
+++ b/src/utils/chromsweep/chromsweep.h
@@ -30,10 +30,10 @@ public:
// A is the query and B is the database
// constructor using existing BedFile pointers
- ChromSweep(BedFile *bedA, BedFile *bedB, bool sameStrand = false, bool diffStrand = false);
+ ChromSweep(BedFile *query, BedFile *db, bool sameStrand = false, bool diffStrand = false, bool printHeader = false);
// constructor using filenames
- ChromSweep(string &bedAFile, string &bedBFile);
+ ChromSweep(string &queryFile, string &dbFile);
// destructor
~ChromSweep(void);
@@ -57,7 +57,7 @@ public:
private:
// instances of a bed file class.
- BedFile *_bedA, *_bedB;
+ BedFile *_query, *_db;
// do we care about strandedness.
bool _sameStrand, _diffStrand;
// a cache of still active features from the database file
@@ -73,10 +73,6 @@ private:
BED _curr_qy, _curr_db;
// a cache of the current chrom from the query. used to handle chrom changes.
string _curr_chrom;
- // the current line status in the database and query files
- BedLineStatus _qy_status, _db_status;
- // the current line numbers in the database and query files
- int _qy_lineNum, _db_lineNum;
// private methods.
private:
diff --git a/src/windowBed/windowBed.cpp b/src/windowBed/windowBed.cpp
index b93ea4ae..63d97745 100644
--- a/src/windowBed/windowBed.cpp
+++ b/src/windowBed/windowBed.cpp
@@ -129,9 +129,7 @@ void BedWindow::WindowIntersectBed() {
_bedB->loadBedFileIntoMap();
BED a, nullBed;
- int lineNum = 0; // current input line number
- BedLineStatus bedStatus;
- vector<BED> hits; // vector of potential hits
+ vector<BED> hits;
hits.reserve(100);
_bedA->Open();
@@ -139,8 +137,8 @@ void BedWindow::WindowIntersectBed() {
if (_printHeader == true) {
_bedA->PrintHeader();
}
- while ((bedStatus = _bedA->GetNextBed(a, lineNum)) != BED_INVALID) {
- if (bedStatus == BED_VALID) {
+ while (_bedA->GetNextBed(a)) {
+ if (_bedA->_status == BED_VALID) {
FindWindowOverlaps(a, hits);
hits.clear();
a = nullBed;
--
GitLab