[med-svn] [Git][med-team/snap-aligner][upstream] New upstream version 2.0.3+dfsg
Andreas Tille (@tille)
gitlab at salsa.debian.org
Thu Aug 24 10:22:30 BST 2023
Andreas Tille pushed to branch upstream at Debian Med / snap-aligner
Commits:
fd2267ee by Andreas Tille at 2023-08-24T11:08:21+02:00
New upstream version 2.0.3+dfsg
- - - - -
8 changed files:
- SNAPLib/Bam.cpp
- SNAPLib/BaseAligner.cpp
- SNAPLib/CommandProcessor.cpp
- SNAPLib/DataReader.cpp
- SNAPLib/DataReader.h
- SNAPLib/IntersectingPairedEndAligner.cpp
- SNAPLib/SingleAligner.cpp
- SNAPLib/SortedDataWriter.cpp
Changes:
=====================================
SNAPLib/Bam.cpp
=====================================
@@ -1025,6 +1025,7 @@ BAMFormat::writeHeader(
}
*headerActualSize = cursor;
+
return true;
}
=====================================
SNAPLib/BaseAligner.cpp
=====================================
@@ -693,7 +693,7 @@ Return Value:
//
// And finally, try scoring.
//
- if (score(
+ bool success = score(
false,
read,
primaryResult,
@@ -706,18 +706,21 @@ Return Value:
maxCandidatesForAffineGapBufferSize,
nCandidatesForAffineGap,
candidatesForAffineGap,
- useHamming)) {
+ useHamming);
+ if (overflowedSecondaryResultsBuffer) {
+ return false;
+ }
+
+ if (success) {
#ifdef _DEBUG
if (_DumpAlignments) printf("Final result score %d MAPQ %d at %s:%llu\n", primaryResult->score, primaryResult->mapq,
genome->getContigAtLocation(primaryResult->location)->name, primaryResult->location - genome->getContigAtLocation(primaryResult->location)->beginningLocation);
#endif // _DEBUG
- if (overflowedSecondaryResultsBuffer) {
- return false;
- }
+
finalizeSecondaryResults(read[FORWARD], primaryResult, nSecondaryResults, secondaryResults, maxSecondaryResults, maxEditDistanceForSecondaryResults, primaryResult->score);
return true;
- } // If score says we have a difinitive answer
+ } // If score says we have a definitive answer
} // If we applied a seed, and so something's changed.
} // While we're still applying seeds
=====================================
SNAPLib/CommandProcessor.cpp
=====================================
@@ -36,7 +36,7 @@ Revision History:
#include "Compat.h"
#include "HitDepth.h"
-const char *SNAP_VERSION = "2.0.2";
+const char *SNAP_VERSION = "2.0.3";
static void usage()
{
=====================================
SNAPLib/DataReader.cpp
=====================================
@@ -60,6 +60,8 @@ public:
virtual bool getData(char** o_buffer, _int64* o_validBytes, _int64* o_startBytes = NULL);
+ virtual void dumpState();
+
virtual void advance(_int64 bytes);
virtual void nextBatch();
@@ -165,6 +167,16 @@ protected:
const size_t bufferSize;
};
+void
+ReadBasedDataReader::dumpState()
+{
+ WriteErrorMessage("ReadBasedDataReader at 0x%llx state:\n", this);
+ WriteErrorMessage("\theaderBufferSize %lld, headerExtraSize %lld, amountAdvancedThroughUnderlyingStoreByUs %lld, nHeaderBuffersAllocated %d, hitEOFReadingHeader %d, bufferSize %lld\n",
+ headerBufferSize, headerExtraSize, amountAdvancedThroughUnderlyingStoreByUs, nHeaderBuffersAllocated, hitEOFReadingHeader, bufferSize);
+ WriteErrorMessage("\tnBuffers %d, headerBuffersOutstanding, %d, startedReadingHeader %d, extraBytes %lld, overflowBytes %lld, nextBatchID %d, nextBufferForReader %d, nextBufferForConsumer %d, lastBufferForConsumer %d\n",
+ nBuffers, headerBuffersOutstanding, startedReadingHeader, extraBytes, overflowBytes, nextBatchID, nextBufferForReader, nextBufferForConsumer, lastBufferForConsumer);
+}
+
ReadBasedDataReader::ReadBasedDataReader(
unsigned i_nBuffers,
_int64 i_overflowBytes,
@@ -186,12 +198,20 @@ ReadBasedDataReader::ReadBasedDataReader(
_ASSERT(extraFactor >= 0 && i_nBuffers > 0);
bufferInfo = new BufferInfo[maxBuffers];
extraBytes = max((_int64) 0, (_int64) ((bufferSize + overflowBytes) * extraFactor));
+
+ if (bufferSize <= 2 * (size_t)overflowBytes) {
+ WriteErrorMessage("ReadBasedDataReader::ReadBasedDataReader: the buffer size %lld isn't more than twice the overflow size %lld, so we will fail to make progres. This is a code bug, please create a github issue.\n",
+ bufferSize, overflowBytes);
+ soft_exit(1);
+ }
+
char* allocated = (char*) BigReserve(maxBuffers * (bufferSize + extraBytes + overflowBytes));
BigCommit(allocated, nBuffers * (bufferSize + extraBytes + overflowBytes));
if (NULL == allocated) {
WriteErrorMessage("ReadBasedDataReader: unable to allocate IO buffer\n");
soft_exit(1);
}
+
for (unsigned i = 0 ; i < nBuffers; i++) {
bufferInfo[i].buffer = allocated;
allocated += bufferSize + overflowBytes;
@@ -206,6 +226,7 @@ ReadBasedDataReader::ReadBasedDataReader(
bufferInfo[i].holds = 0;
bufferInfo[i].headerBuffer = false;
}
+
nextBatchID = 1;
nextBufferForConsumer = -1;
@@ -1331,6 +1352,8 @@ public:
return fileName;
}
+ virtual void dumpState();
+
protected:
// must hold the lock to call
@@ -1354,11 +1377,28 @@ protected:
};
+void
+AsyncFileDataReader::dumpState()
+{
+ WriteErrorMessage("AsyncFileDataReader (0x%llx) state:\n", this);
+ WriteErrorMessage("\tfileName %s, fileSize %lld, readOffset %lld, endingOffset %lld\n",
+ fileName, fileSize, readOffset, endingOffset);
+
+ ReadBasedDataReader::dumpState();
+}
+
AsyncFileDataReader::AsyncFileDataReader(unsigned i_nBuffers, _int64 i_overflowBytes, double extraFactor, size_t bufferSpace) :
ReadBasedDataReader(i_nBuffers, i_overflowBytes, extraFactor, bufferSpace), fileName(NULL), asyncFile(NULL), endingOffset(0)
{
readOffset = 0;
bufferReaders = (AsyncFile::Reader**)malloc(sizeof(AsyncFile::Reader*) * maxBuffers);
+
+ if (NULL == bufferReaders) {
+ WriteErrorMessage("AsyncFileDataReader::AsyncFileDataReader(): malloc(%d) failed\n", sizeof(AsyncFile::Reader*) * maxBuffers);
+ soft_exit(1);
+ return; // Just to avoid the compiler warning
+ }
+
for (unsigned i = 0; i < i_nBuffers; i++) {
bufferReaders[i] = NULL;
}
=====================================
SNAPLib/DataReader.h
=====================================
@@ -145,6 +145,11 @@ public:
// timing for performance tuning (in nanos)
static volatile _int64 ReadWaitTime;
static volatile _int64 ReleaseWaitTime;
+
+ //
+ // debugging
+ //
+ virtual void dumpState() {} // Override in a subclass if needed
};
class DataSupplier
=====================================
SNAPLib/IntersectingPairedEndAligner.cpp
=====================================
@@ -879,10 +879,11 @@ IntersectingPairedEndAligner::alignLandauVishkin(
scoreLimit = computeScoreLimit(nonALTAlignment, &scoresForAllAlignments, &scoresForNonAltAlignments, __max(mate->largestBigIndelDetected, __min(candidate->largestBigIndelDetected, fewerEndScore)));
_ASSERT(genomeLocationIsWithin(mate->readWithMoreHitsGenomeLocation, candidate->readWithFewerHitsGenomeLocation, maxSpacing));
+
//
- // Exclude it if it's strictly smaller than minSpacing; hence, minSpacing -1.
+ // Exclude it if it's strictly smaller than minSpacing; hence, minSpacing - 1.
//
- if (!genomeLocationIsWithin(mate->readWithMoreHitsGenomeLocation, candidate->readWithFewerHitsGenomeLocation, minSpacing -1) && ((mate->bestPossibleScore <= scoreLimit - fewerEndScore))) {
+ if (!genomeLocationIsWithin(mate->readWithMoreHitsGenomeLocation, candidate->readWithFewerHitsGenomeLocation, minSpacing - 1) && ((mate->bestPossibleScore <= scoreLimit - fewerEndScore))) {
//
// It's within the range and not necessarily too poor of a match. Consider it.
//
@@ -1063,7 +1064,7 @@ IntersectingPairedEndAligner::alignLandauVishkin(
}
(*nLVCandidatesForAffineGap)++;
- }
+ } // if not eliminatedByMerge
if (nonALTAlignment) {
scoresForNonAltAlignments.updateBestHitIfNeeded(pairScore, pairAGScore, pairProbability, fewerEndScore, readWithMoreHits, fewerEndGenomeLocationOffset, candidate, mate);
@@ -3369,7 +3370,7 @@ IntersectingPairedEndAligner::scoreLocation(
*totalIndelsLV = totalIndels1 + totalIndels2;
} else {
*score = ScoreAboveLimit;
- *agScore = -1;
+ *agScore = ScoreAboveLimit;
*matchProbability = 0.0;
}
=====================================
SNAPLib/SingleAligner.cpp
=====================================
@@ -247,7 +247,7 @@ SingleAlignerContext::runIterationThreadImpl(Read *& read)
}
#endif
SingleAlignmentResult firstALTResult;
- while (!aligner->AlignRead(read, alignmentResults, &firstALTResult, maxSecondaryAlignmentAdditionalEditDistance, alignmentResultBufferCount - 1, &nSecondaryResults, maxSecondaryAlignments, alignmentResults + 1, 0, NULL, NULL, useSoftClipping)) {
+ while (!aligner->AlignRead(read, alignmentResults, &firstALTResult, maxSecondaryAlignmentAdditionalEditDistance, alignmentResultBufferCount - 1, &nSecondaryResults, maxSecondaryAlignments, alignmentResults + 1, 0, NULL, NULL)) {
//
// Out of secondary alignment buffer. Reallocate.
//
=====================================
SNAPLib/SortedDataWriter.cpp
=====================================
@@ -1260,7 +1260,10 @@ SortedDataFilterSupplier::mergeSort()
{
i->reader = readerSupplier->getDataReader(1, MAX_READ_LENGTH * 8, 0.0,
__min(1UL << 23, __max(1UL << 17, bufferSpace / blocks.size()))); // 128kB to 8MB buffer space per block
- i->reader->init(tempFileName);
+ if (!i->reader->init(tempFileName)) {
+ WriteErrorMessage("SortedDataFilterSupplier::mergeSort: reader->init(%s) failed\n", tempFileName);
+ soft_exit(1);
+ }
i->reader->reinit(i->start, i->bytes);
}
}
@@ -1274,8 +1277,11 @@ SortedDataFilterSupplier::mergeSort()
DataReader* headerReader;
if (blocks[0].dataReaderIsBuffer)
{
- headerReader = readerSupplier->getDataReader(1, MAX_READ_LENGTH * 8, 0.0, headerSize + 4096);
- headerReader->init(tempFileName);
+ headerReader = readerSupplier->getDataReader(1, 0, 0.0, 0);
+ if (!headerReader->init(tempFileName)) {
+ WriteErrorMessage("SortedDataFilterSupplier::mergeSort: reader->init(%s) failed for headerReader\n", tempFileName);
+ soft_exit(1);
+ }
} else {
headerReader = blocks[0].reader;
}
@@ -1289,10 +1295,12 @@ SortedDataFilterSupplier::mergeSort()
if ((!headerReader->getData(&rbuffer, &rbytes)) || rbytes == 0) {
headerReader->nextBatch();
if (!headerReader->getData(&rbuffer, &rbytes)) {
- WriteErrorMessage( "read header failed\n");
+ WriteErrorMessage( "read header failed, left %lld, headerSize %lld\n", left, headerSize);
+ headerReader->dumpState();
soft_exit(1);
}
}
+
if ((! writer->getBuffer(&wbuffer, &wbytes)) || wbytes == 0) {
writer->nextBatch();
if (! writer->getBuffer(&wbuffer, &wbytes)) {
@@ -1303,6 +1311,7 @@ SortedDataFilterSupplier::mergeSort()
size_t xfer = min(left, min((size_t) rbytes, wbytes));
_ASSERT(xfer > 0 && xfer <= UINT32_MAX);
memcpy(wbuffer, rbuffer, xfer);
+
headerReader->advance(xfer);
writer->advance((unsigned) xfer);
left -= xfer;
View it on GitLab: https://salsa.debian.org/med-team/snap-aligner/-/commit/fd2267ee08146eeac05c98010e6c8bdefc5e2f66
--
View it on GitLab: https://salsa.debian.org/med-team/snap-aligner/-/commit/fd2267ee08146eeac05c98010e6c8bdefc5e2f66
You're receiving this email because of your account on salsa.debian.org.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://alioth-lists.debian.net/pipermail/debian-med-commit/attachments/20230824/9def1dd4/attachment-0001.htm>
More information about the debian-med-commit
mailing list