changes related to parameters & default files - v3.4.1

Changelog

@@ -1,3 +1,8 @@
+2018-11-12  Can Alkan  <calkan@gmail.com>
+	* Default output file is now (seqFile1)-output.sam(.gz) & default nohit file is (seqFile1)-output.nohit.fastq(.gz)
+	* No-hit FASTQ output is new also compressed when --outcomp parameter is used.
+	* mrsFAST now accepts -t as a short form for --threads.
+
 2014-03-31  Faraz Hach  <fhach@sfu.ca>
 	* Added with-sse4 option to Makefile
 	* Bug reported by viktor in dc mode is fixed.

CommandLineParser.h

@@ -38,5 +38,5 @@
 
 int parseCommandLine (int argc, char *argv[]);
 void finalizeCommandParser();
-
+void printHelp(void);
 #endif

Common.c

@@ -61,8 +61,8 @@ FILE *fileOpen(char *fileName, char *mode)
 	fp = fopen (fileName, mode);
 	if (fp == NULL)
 	{
-		fprintf(stdout, "Error: Cannot Open the file %s\n", fileName);
-		fflush(stdout);
+		fprintf(stderr, "Error: Cannot Open the file %s\n", fileName);
+		fflush(stderr);
 		exit(EXIT_FAILURE);
 	}
 	return fp;
@@ -74,8 +74,8 @@ gzFile fileOpenGZ(char *fileName, char *mode)
 	gzfp = gzopen (fileName, mode);
 	if (gzfp == Z_NULL)
 	{
-		fprintf(stdout, "Error: Cannot Open the file %s\n", fileName);
-		fflush(stdout);
+		fprintf(stderr, "Error: Cannot Open the file %s\n", fileName);
+		fflush(stderr);
 		exit(EXIT_FAILURE);
 	}
 	return gzfp;

Common.h

@@ -77,6 +77,7 @@ extern int				bestMappingMode;
 extern int				SNPMode;
 extern int				seqCompressed;
 extern int				outCompressed;
+extern int				isCloud;
 extern int				cropSize;
 extern int				tailCropSize;
 extern int				progressRep;

HashTable.c

@@ -232,8 +232,8 @@ int generateHashTable(char *fileName, char *indexName)
 	if (!initLoadingRefGenome(fileName, genomeMetaInfo, &genomeMetaInfoLength))
 		return 0;		
 	initSavingIHashTable(indexName, genomeMetaInfo, genomeMetaInfoLength);
-	fprintf(stdout, "Generating Index from %s", fileName);
-	fflush(stdout);
+	fprintf(stderr, "Generating Index from %s", fileName);
+	fflush(stderr);
 
 	do
 	{
@@ -244,14 +244,14 @@ int generateHashTable(char *fileName, char *indexName)
 
 		if ( strcmp(prev, refGenName) != 0)
 		{
-			fprintf(stdout, "\n - %s ", refGenName);
-			fflush(stdout);
+			fprintf(stderr, "\n - %s ", refGenName);
+			fflush(stderr);
 			sprintf(prev, "%s", refGenName);
 		}
 		else
 		{
-			fprintf(stdout, ".");
-			fflush(stdout);
+			fprintf(stderr, ".");
+			fflush(stderr);
 		}
 
 		c = refGen;
@@ -296,7 +296,7 @@ int generateHashTable(char *fileName, char *indexName)
 	finalizeLoadingRefGenome();
 	finalizeSavingIHashTable();
 
-	fprintf(stdout, "\nDONE in %0.2fs!\n", (getTime()-startTime));
+	fprintf(stderr, "\nDONE in %0.2fs!\n", (getTime()-startTime));
 	return 1;
 }
 /**********************************************/
@@ -315,12 +315,12 @@ int checkHashTable(char *fileName)
 	tmp = fread(&magicNumber, sizeof(magicNumber), 1, _ih_fp);
 	if (magicNumber == 1)
 	{
-		fprintf(stdout, "Error: Please use version 1.2.6.4 in bisulfite mode.\n");
+		fprintf(stderr, "Error: Please use version 1.2.6.4 in bisulfite mode.\n");
 		return 0;
 	}
 	else if (magicNumber == 0)
 	{
-		fprintf(stdout, "Error: Please use version 2.x.x.x or upgrade your index.\n");
+		fprintf(stderr, "Error: Please use version 2.x.x.x or upgrade your index.\n");
 		return 0;
 	}
 

Makefile

@@ -1,16 +1,16 @@
-MRSFAST_VERSION := "3.4.0"
+MRSFAST_VERSION := "3.4.8"
 BUILD_DATE := "$(shell date)"
 
 all: OPTIMIZE_FLAGS build
 debug: DEBUG_FLAGS build
 profile: PROFILE_FLAGS build
-build: clean_executables SSE_FLAGS mrsfast snp_indexer clean_objects
+build: clean_executables mrsfast snp_indexer clean_objects
 
 LDFLAGS=#-static
-LIBS=-lz -lm -pthread -lpthread
-CFLAGS=-fno-pic -DMRSFAST_VERSION=\"$(MRSFAST_VERSION)\" -DBUILD_DATE=\"$(BUILD_DATE)\"
+LIBS=-lz -lm -pthread -lpthread  -DSSE4=1 -msse4.2
+CFLAGS=-DMRSFAST_VERSION=\"$(MRSFAST_VERSION)\" -DBUILD_DATE=\"$(BUILD_DATE)\" -DSSE4=1 -msse4.2
 
-objects=baseFAST.o Sort.o MrsFAST.o Common.o CommandLineParser.o RefGenome.o HashTable.o Reads.o Output.o SNPReader.o  HELP.o
+objects=baseFAST.o Sort.o MrsFAST.o Common.o CommandLineParser.o RefGenome.o HashTable.o Reads.o Output.o SNPReader.o  
 
 mrsfast: clean_executables $(objects)
 ifeq ($(shell uname -s),Linux)
@@ -38,15 +38,6 @@ clean_executables:
 	@rm -f mrsfast
 	@rm -f snp_indexer
 
-HELP.o:
-	@groff -Tascii -man ./HELP.man > HELP
-ifeq ($(shell uname -s),Linux)
-	@ld -r -b binary -o HELP.o HELP
-else
-	@touch HELPstub.c
-	gcc -o HELPstub.o -c HELPstub.c
-	ld -r -o HELP.o -sectcreate binary HELP HELP HELPstub.o
-endif
 
 DEBUG_FLAGS:
 	$(eval CFLAGS = $(CFLAGS) -ggdb)
@@ -59,23 +50,3 @@ PROFILE_FLAGS:
 		$(eval CFLAGS = $(CFLAGS) -pg -g)
 	$(eval LIBS = $(LIBS) -pg -g)
 
-SSE_FLAGS:
-ifeq ($(shell uname -s),Linux)
-ifeq ($(with-sse4),no)
-		$(shell echo "-DSSE4=0")
-else
-        	$(eval CFLAGS = $(CFLAGS) \
-        	$(shell gv=`gcc -dumpversion`; \
-            	    sc=`grep -c "sse4" /proc/cpuinfo`; \
-                	echo $$sc.$$gv | awk -F. '{if($$1>0 && $$2>=4 && $$3>=4) print "-DSSE4=1 -msse4.2"; else print "-DSSE4=0"}'))
-endif
-else
-ifeq ($(with-sse4),no)
-		$(shell echo "-DSSE4=0")
-else
-        $(eval CFLAGS = $(CFLAGS) \
-        $(shell gv=`gcc -dumpversion`; \
-                sc=`sysctl -n machdep.cpu.features | grep -c "SSE4"` ;\
-                echo $$sc.$$gv | awk -F. '{if($$1>0 && $$2>=4 && $$3>=4) print "-DSSE4=1 -msse4.2"; else print "-DSSE4=0"}'))
-endif
-endif

MrsFAST.c

@@ -2692,34 +2692,34 @@ void outputBestPairedEnd()
 				f1 = 1 + 4 + 8 + 64;
 				f2 = 1 + 4 + 8 + 128;
 				optSize1 = optSize2 = 0;
-				//fprintf(stdout, "unset\n");
+				//fprintf(stderr, "unset\n");
 				break;
 			case first_mate:
 				f1 = 1 + 8 + 64 + ((_msf_bestMappingPE[i].dir1 == -1) ?16 :0);
 				f2 = 1 + 4 + 128 + ((_msf_bestMappingPE[i].dir1 == -1) ?32 :0);
 				optSize2 = 0;
-				//fprintf(stdout, "1stmate\n");
+				//fprintf(stderr, "1stmate\n");
 				break;
 			case second_mate:
 				f1 = 1 + 4 + 64 + ((_msf_bestMappingPE[i].dir2 == -1) ?32 :0);
 				f2 = 1 + 8 + 128 + ((_msf_bestMappingPE[i].dir2 == -1) ?16 :0);
-				//fprintf(stdout, "2ndmate\n");
+				//fprintf(stderr, "2ndmate\n");
 				optSize1 = 0;
 				break;
 			case trans_loc:
 				f1 = 1 + 64 + ((_msf_bestMappingPE[i].dir1 == -1) ?16 :0) + ((_msf_bestMappingPE[i].dir2 == -1) ?32 :0);
 				f2 = 1 + 128 + ((_msf_bestMappingPE[i].dir2 == -1) ?16 :0) + ((_msf_bestMappingPE[i].dir1 == -1) ?32 :0);
-				//fprintf(stdout, "transLoc %d %d\n", f1, f2);
+				//fprintf(stderr, "transLoc %d %d\n", f1, f2);
 				break;
 			case improper:
 				f1 = 1 + 64 + ((_msf_bestMappingPE[i].dir1 == -1) ?16 :0) + ((_msf_bestMappingPE[i].dir2 == -1) ?32 :0);
 				f2 = 1 + 128 + ((_msf_bestMappingPE[i].dir2 == -1) ?16 :0) + ((_msf_bestMappingPE[i].dir1 == -1) ?32 :0);
-				//fprintf(stdout, "improper\n");
+				//fprintf(stderr, "improper\n");
 				break;
 			case proper:
 				f1 = 1 + 2 + 64 + ((_msf_bestMappingPE[i].dir1 == -1) ?16 :0) + ((_msf_bestMappingPE[i].dir2 == -1) ?32 :0);
 				f2 = 1 + 2 + 128 + ((_msf_bestMappingPE[i].dir2 == -1) ?16 :0) + ((_msf_bestMappingPE[i].dir1 == -1) ?32 :0);
-				//fprintf(stdout, "proper\n");
+				//fprintf(stderr, "proper\n");
 				break;
 		}
 		//output best
@@ -3415,7 +3415,7 @@ void updateBestPairedEnd()
 		}
 
 
-//fprintf(stdout, "HERE?\n");
+//fprintf(stderr, "HERE?\n");
 /*		pos = 0;
 
 		for (j=0; j<size1; j++)
@@ -3622,7 +3622,7 @@ void outputPairedEndDiscPP()
 			int dist = 0;
 			char event;
 
-			//fprintf(stdout, "%c %c ", dir1, dir2);
+			//fprintf(stderr, "%c %c ", dir1, dir2);
 
 			if (loc1 == loc2)
 			{
@@ -3633,49 +3633,49 @@ void outputPairedEndDiscPP()
 				if ( dir1 == dir2 )
 				{
 					event = 'V';
-					//fprintf(stdout, "Inverstion \n");
+					//fprintf(stderr, "Inverstion \n");
 				}
 				else
 				{
 					if (loc1 < loc2)
 					{
 
-						//fprintf(stdout, "< %d ", loc2-loc1-SEQ_LENGTH);
+						//fprintf(stderr, "< %d ", loc2-loc1-SEQ_LENGTH);
 
 						if (dir1 == 'R' && dir2 == 'F')
 						{
 							event = 'E';
 
-							//fprintf(stdout, "Everted \n");
+							//fprintf(stderr, "Everted \n");
 						}
 						else if ( loc2 - loc1 >= maxPairEndedDiscordantDistance )
 						{
 							event = 'D';
-							//fprintf(stdout, "Deletion \n");
+							//fprintf(stderr, "Deletion \n");
 						}
 						else
 						{
 							event = 'I';
-							//fprintf(stdout, "Insertion \n");
+							//fprintf(stderr, "Insertion \n");
 						}
 					}
 					else if (loc2 < loc1)
 					{
-						//fprintf(stdout, "> %d ", loc1-loc2-SEQ_LENGTH);
+						//fprintf(stderr, "> %d ", loc1-loc2-SEQ_LENGTH);
 						if (dir2 == 'R' && dir1 == 'F')
 						{
 							event = 'E';
-							//fprintf(stdout, "Everted \n");
+							//fprintf(stderr, "Everted \n");
 						}
 						else if ( loc1 - loc2 >= maxPairEndedDiscordantDistance )
 						{
 							event = 'D';
-							//fprintf(stdout, "Deletion \n");
+							//fprintf(stderr, "Deletion \n");
 						}
 						else
 						{
 							event = 'I';
-							//fprintf(stdout, "Insertion \n");
+							//fprintf(stderr, "Insertion \n");
 						}
 					}
 				}
@@ -4080,7 +4080,7 @@ void calculateConcordantDistances()
 
 	minPairEndedDistance = (int)(mu - 3*sigma);
 	maxPairEndedDistance = (int)(mu + 3*sigma);
-	//fprintf(stdout, "cnt %d  mu %lf  sig %lf  min %d  max %d\n", cnt, mu, sigma, minPairEndedDistance, maxPairEndedDistance);
+	//fprintf(stderr, "cnt %d  mu %lf  sig %lf  min %d  max %d\n", cnt, mu, sigma, minPairEndedDistance, maxPairEndedDistance);
 
 	FILE *out = fileOpen(concordantStatOutput, "w");
 	fprintf(out, "TotalNumbReads: 2*%d\n", (_msf_seqListSize / 2));

Output.c

@@ -163,7 +163,11 @@ int initOutput ( char *fileName, int compressed)
 	if (compressed)
 	{
 		char newFileName[strlen(mappingOutputPath)+strlen(fileName)+4];
-		sprintf(newFileName, "%s%s.sam.gz", mappingOutputPath, fileName);
+		if (!isCloud)
+		  sprintf(newFileName, "%s%s.sam.gz", mappingOutputPath, fileName);
+		else
+		  sprintf(newFileName, "%s.sam.gz", fileName);
+		fprintf(stderr, "mappingOutputPath: %s\nfileName: %s\nnewFileName: %s\n", mappingOutputPath, fileName, newFileName);
 		_out_gzfp = fileOpenGZ(newFileName, "w1f");
 		if (_out_gzfp == Z_NULL)
 		{
@@ -187,11 +191,28 @@ int initOutput ( char *fileName, int compressed)
 		}
 		else
 		{
-			//sprintf(newFileName, "%s%s.sam", mappingOutputPath, fileName);
-			sprintf(newFileName, "%s%s", mappingOutputPath, fileName);
+		  if (!isCloud){	   
+		          int fnlen = strlen(fileName);
+			  if ( ! (fileName[fnlen-1]=='m' && fileName[fnlen-2]=='a' && fileName[fnlen-3]=='s' ) )
+			    sprintf(newFileName, "%s%s.sam", mappingOutputPath, fileName);
+			  else
+			    sprintf(newFileName, "%s%s", mappingOutputPath, fileName);
+		  }
+		  else{
+		          int fnlen = strlen(fileName);
+			  if ( ! (fileName[fnlen-1]=='m' && fileName[fnlen-2]=='a' && fileName[fnlen-3]=='s' ) )
+			    sprintf(newFileName, "%s.sam", fileName);
+			  else
+			    sprintf(newFileName, "%s", fileName);
+			//sprintf(newFileName, "%s%s", mappingOutputPath, fileName);
+		  }
 		}
 
-		_out_fp = fileOpen(newFileName, "w");
+		if (!strstr(newFileName, "/dev/stdout"))
+		  _out_fp = fileOpen(newFileName, "w");
+		else
+		  _out_fp = stdout;
+
 		if (_out_fp == NULL)
 		{
 			return 0;