Removed deadlock in the -multisample-fasta mode.

Author: agudys

README.md

@@ -148,6 +148,8 @@ Parameters:
 * `common_table` (output) - file containing table with common k-mer counts.
 * `-multisample-fasta` / `-from-kmers` / `-from-minhash` - see `build` mode for details.
 * `-sparse` - stores output matrix in a sparse form,
+* `-above <a_th>` - retains elements larger then <a_th>,
+* `-below <b_th>` - retains elements smaller then <b_th>,
 * `-t <threads>` - number of threads (default: number of available cores).
  
 ### Single sample against the database:

src/input_file.cpp

@@ -189,11 +189,6 @@ bool GenomeInputFile::loadNext(
 	std::string& sampleName
 ) {
 
-	// no more sequences in multifasta
-	if (multifastaIndex == chromosomes.size()) {
-		return false;
-	}
-
 	std::shared_ptr<MinHashFilter> minhashFilter = dynamic_pointer_cast<MinHashFilter>(filter);
 
 	if (!minhashFilter) {
@@ -219,7 +214,8 @@ bool GenomeInputFile::loadNext(
 
 	++multifastaIndex;
 
-	return true;
+	// no more sequences in multifasta
+	return multifastaIndex < chromosomes.size();
 }
 
 

src/loader_ex.cpp

@@ -191,11 +191,12 @@ void LoaderEx::multifastaReaderJob() {
 		// initialize multifasta file
 		if (genomicFile->initMultiFasta()) {
 
-			LOG_DEBUG << "multifasta initialized " << endl ;
+			LOG_DEBUG << "multifasta initialized: " << inputTask->fileId + 1 << endl ;
 
 			while (true) {
-
+				LOG_DEBUG << "wait for buf for sample " << sample_id + 1 << endl;
 				this->queues.freeBuffers.Pop(bufferId); // wait for free buffer
+				LOG_DEBUG << "acquired buf " << bufferId << " for sample " << sample_id + 1 << endl;
 
 				auto sampleTask = make_shared<SampleTask>(
 					sample_id,
@@ -207,17 +208,17 @@ void LoaderEx::multifastaReaderJob() {
 					kmersCollections[bufferId], positionsCollections[bufferId],
 					sampleTask->kmers, sampleTask->kmersCount, sampleTask->kmerLength, sampleTask->fraction, sampleTask->sampleName);
 
-				// no more samples
-				if (!ok) {
-					break;
-				}
-
 				++sample_id;
 				++bufferRefCounters[bufferId];
 				queues.output.Push((int)sampleTask->id, sampleTask);
 				++count;
 
-				LOG_DEBUG << "(sample " << sampleTask->id + 1 << ") -> output queue, buf: " << bufferId << std::endl ;
+				LOG_DEBUG << "(sample " << sampleTask->id + 1 << ") -> output queue, buf: " << bufferId << std::endl;
+
+				// no more samples
+				if (!ok) {
+					break;
+				}
 			}
 		}
 

src/version.h

@@ -1,12 +1,15 @@
 #pragma once
 
-#define VERSION "1.11.0"
-#define DATE "27.02.2023"
+#define VERSION "1.11.1"
+#define DATE "07.03.2023"
 
 /* 
 
 Version history
 
+1.11.1 (07.03.2023)
+- Removed deadlock in the -multisample-fasta mode.
+
 1.11.0 (27.02.2023)
 - Added -below and -above thresholds for all2all and new2all modes.