bio-velvet_underground 0.1.0 → 0.2.0

data/ext/src/src/run.c

@@ -39,7 +39,7 @@ static void printUsage()
 	printf("\thash_length\t: EITHER an odd integer (if even, it will be decremented) <= %i (if above, will be reduced)\n", MAXKMERLENGTH);
 	printf("\t\t\t: OR: m,M,s where m and M are odd integers (if not, they will be decremented) with m < M <= %i (if above, will be reduced)\n", MAXKMERLENGTH);
 	puts("\t\t\t\tand s is a step (even number). Velvet will then hash from k=m to k=M with a step of s");
-	puts("\tfilename\t: path to sequence file or - for standard input");
+	puts("\tfilename\t: path to sequence file or - for standard input");	
 	puts("");
 	puts("File format options:");
 	puts("\t-fasta\t-fastq\t-raw\t-fasta.gz\t-fastq.gz\t-raw.gz\t-sam\t-bam\t-fmtAuto");
@@ -52,7 +52,7 @@ static void printUsage()
 	puts("Read type options:");
 	puts("\t-short\t-shortPaired");
 #if CATEGORIES <= 5
-	Category cat;
+	Category cat; 
 	for (cat = 2; cat <= CATEGORIES; cat++)
 	    printf("\t-short%i\t-shortPaired%i\n", cat, cat);
 #else
@@ -93,7 +93,7 @@ static void printUsage()
 	puts("\t\t[Both files are picked up by graph, so please leave them there]");
 }
 
-int velveth(int argc, char **argv)
+int main(int argc, char **argv)
 {
 	ReadSet *allSequences = NULL;
 	SplayTable *splayTable;
@@ -161,18 +161,18 @@ int velveth(int argc, char **argv)
 		    ("Velvet can't handle k-mers as long as %i! We'll stick to %i if you don't mind.\n",
 		     hashLength, MAXKMERLENGTH);
 		hashLength = MAXKMERLENGTH;
-	}
+	} 
 	if (hashLength <= 0) {
 		velvetLog("Invalid hash length: %s\n", argv[2]);
 		printUsage();
 		return 0;
-	}
+	} 
 	if (hashLength % 2 == 0) {
 		velvetLog
 		    ("Velvet can't work with even length k-mers, such as %i. We'll use %i instead, if you don't mind.\n",
 		     hashLength, hashLength - 1);
 		hashLength--;
-	}
+	} 
 
 	if (multiple_kmers) {
 		if (hashLengthMax > MAXKMERLENGTH + 1) {
@@ -180,12 +180,12 @@ int velveth(int argc, char **argv)
 			    ("Velvet can't handle k-mers as long as %i! We'll stick to %i if you don't mind.\n",
 			     hashLengthMax, MAXKMERLENGTH + 1);
 			hashLengthMax = MAXKMERLENGTH + 1;
-		}
+		} 
 		if (hashLengthMax <= hashLength) {
 			velvetLog("hashLengthMin < hashLengthMax is required %s", argv[2]);
 			printUsage();
 			return 0;
-		}
+		} 
 
 		if (hashLengthStep <= 0) {
 			velvetLog("Non-positive hash length! Setting it to 2\n");
@@ -215,7 +215,7 @@ int velveth(int argc, char **argv)
 			sprintf(buf,"%s_%d",argv[1],h);
 			directory = mallocOrExit(strlen(buf) + 100, char);
 			strcpy(directory,buf);
-		} else
+		} else 
 			directory = argv[1];
 
 		filename = mallocOrExit(strlen(directory) + 100, char);

data/ext/src/src/run2.c

@@ -68,8 +68,6 @@ static void printUsage()
 	puts("\t-paired_exp_fraction <double>\t: remove all the paired end connections which less than the specified fraction of the expected count (default: 0.1)");
 	puts("\t-shortMatePaired* <yes|no>\t: for mate-pair libraries, indicate that the library might be contaminated with paired-end reads (default no)");
 	puts("\t-conserveLong <yes|no>\t\t: preserve sequences with long reads in them (default no)");
-	puts("\t-clip_tips <yes|no>\t\t: do tip clipping on pre-graph (default yes)");
-	puts("\t-tour_bus <yes|no>\t\t: apply the tour bus algorithm (default yes)");
 	puts("");
 	puts("Output:");
 	puts("\tdirectory/contigs.fa\t\t: fasta file of contigs longer than twice hash length");
@@ -78,7 +76,7 @@ static void printUsage()
 	puts("\tdirectory/velvet_asm.afg\t: (if requested) AMOS compatible assembly file");
 }
 
-int velvetg(int argc, char **argv)
+int main(int argc, char **argv)
 {
 	ReadSet *sequences = NULL;
 	RoadMapArray *rdmaps;
@@ -88,9 +86,7 @@ int velvetg(int argc, char **argv)
 	    *preGraphFilename, *seqFilename, *roadmapFilename,
 	    *lowCovContigsFilename, *highCovContigsFilename;
 	double coverageCutoff = -1;
-  boolean doClipTips = true;
-  boolean doTourBus = true;
-  double longCoverageCutoff = -1;
+	double longCoverageCutoff = -1;
 	double maxCoverageCutoff = -1;
 	double expectedCoverage = -1;
 	Coordinate minContigLength = -1;
@@ -168,7 +164,7 @@ int velvetg(int argc, char **argv)
 		return 0;
 	}
 
-	// Memory allocation
+	// Memory allocation 
 	directory = argv[1];
 	graphFilename = mallocOrExit(strlen(directory) + 100, char);
 	connectedGraphFilename = mallocOrExit(strlen(directory) + 100, char);
@@ -185,9 +181,9 @@ int velvetg(int argc, char **argv)
 		if (arg_index >= argc) {
 			velvetLog("Unusual number of arguments!\n");
 			printUsage();
-#ifdef DEBUG
+#ifdef DEBUG 
 			abort();
-#endif
+#endif 
 			exit(1);
 		}
 
@@ -214,9 +210,9 @@ int velvetg(int argc, char **argv)
 			if (insertLength[0] < 0) {
 				velvetLog("Invalid insert length: %lli\n",
 				       (long long) insertLength[0]);
-#ifdef DEBUG
+#ifdef DEBUG 
 				abort();
-#endif
+#endif 
 				exit(1);
 			}
 		} else if (strcmp(arg, "-ins_length_sd") == 0) {
@@ -225,9 +221,9 @@ int velvetg(int argc, char **argv)
 			if (std_dev[0] < 0) {
 				velvetLog("Invalid std deviation: %lli\n",
 				       (long long) std_dev[0]);
-#ifdef DEBUG
+#ifdef DEBUG 
 				abort();
-#endif
+#endif 
 				exit(1);
 			}
 		} else if (strcmp(arg, "-ins_length_long") == 0) {
@@ -242,9 +238,9 @@ int velvetg(int argc, char **argv)
 			cat = (Category) short_var;
 			if (cat < 1 || cat > CATEGORIES) {
 				velvetLog("Unknown option: %s\n", arg);
-#ifdef DEBUG
+#ifdef DEBUG 
 				abort();
-#endif
+#endif 
 				exit(1);
 			}
 			sscanf(argv[arg_index], "%lli", &longlong_var);
@@ -252,9 +248,9 @@ int velvetg(int argc, char **argv)
 			if (insertLength[cat - 1] < 0) {
 				velvetLog("Invalid insert length: %lli\n",
 				       (long long) insertLength[cat - 1]);
-#ifdef DEBUG
+#ifdef DEBUG 
 				abort();
-#endif
+#endif 
 				exit(1);
 			}
 		} else if (strncmp(arg, "-ins_length", 11) == 0) {
@@ -262,9 +258,9 @@ int velvetg(int argc, char **argv)
 			cat = (Category) short_var;
 			if (cat < 1 || cat > CATEGORIES) {
 				velvetLog("Unknown option: %s\n", arg);
-#ifdef DEBUG
+#ifdef DEBUG 
 				abort();
-#endif
+#endif 
 				exit(1);
 			}
 			sscanf(argv[arg_index], "%lli", &longlong_var);
@@ -272,9 +268,9 @@ int velvetg(int argc, char **argv)
 			if (std_dev[cat - 1] < 0) {
 				velvetLog("Invalid std deviation: %lli\n",
 				       (long long) std_dev[cat - 1]);
-#ifdef DEBUG
+#ifdef DEBUG 
 				abort();
-#endif
+#endif 
 				exit(1);
 			}
 		} else if (strcmp(arg, "-read_trkg") == 0) {
@@ -357,12 +353,6 @@ int velvetg(int argc, char **argv)
 				exit(1);
 			}
 			shadows[cat - 1] = (strcmp(argv[arg_index], "yes") == 0);
-    } else if (strcmp(arg,"-clip_tips") == 0){
-      if (strcmp(argv[arg_index], "no") == 0)
-        doClipTips = false;
-    } else if (strcmp(arg,"-tour_bus") == 0){
-      if (strcmp(argv[arg_index], "no") == 0)
-        doTourBus = false;
 		} else if (strcmp(arg, "--help") == 0) {
 			printUsage();
 			return 0;
@@ -462,8 +452,7 @@ int velvetg(int argc, char **argv)
 
 		sequenceLengths =
 		    getSequenceLengths(sequences, getWordLength(graph));
-    if (doTourBus)
-        correctGraph(graph, sequenceLengths, sequences->categories, conserveLong);
+		correctGraph(graph, sequenceLengths, sequences->categories, conserveLong);
 		exportGraph(graphFilename, graph, sequences->tSequences);
 	} else if ((file = fopen(preGraphFilename, "r")) != NULL) {
 		fclose(file);
@@ -479,8 +468,7 @@ int velvetg(int argc, char **argv)
 				   roadmapFilename, readTracking, accelerationBits);
 		sequenceLengths =
 		    getSequenceLengths(sequences, getWordLength(graph));
-    if (doTourBus)
-        correctGraph(graph, sequenceLengths, sequences->categories, conserveLong);
+		correctGraph(graph, sequenceLengths, sequences->categories, conserveLong);
 		exportGraph(graphFilename, graph, sequences->tSequences);
 	} else if ((file = fopen(roadmapFilename, "r")) != NULL) {
 		fclose(file);
@@ -521,7 +509,7 @@ int velvetg(int argc, char **argv)
 		}
 		preGraph = newPreGraph_pg(rdmaps, seqReadInfo);
 		concatenatePreGraph_pg(preGraph);
-		if (!conserveLong && doClipTips)
+		if (!conserveLong)
 		    clipTips_pg(preGraph);
 		exportPreGraph_pg(preGraphFilename, preGraph);
 		destroyPreGraph_pg(preGraph);
@@ -535,14 +523,13 @@ int velvetg(int argc, char **argv)
 				   roadmapFilename, readTracking, accelerationBits);
 		sequenceLengths =
 		    getSequenceLengths(sequences, getWordLength(graph));
-    if (doTourBus)
-        correctGraph(graph, sequenceLengths, sequences->categories, conserveLong);
+		correctGraph(graph, sequenceLengths, sequences->categories, conserveLong);
 		exportGraph(graphFilename, graph, sequences->tSequences);
 	} else {
 		velvetLog("No Roadmap file to build upon! Please run velveth (see manual)\n");
-#ifdef DEBUG
+#ifdef DEBUG 
 		abort();
-#endif
+#endif 
 		exit(1);
 	}
 
@@ -566,11 +553,11 @@ int velvetg(int argc, char **argv)
 			coverageCutoff = expectedCoverage / 2;
 			estimateCutoff = true;
 		}
-	} else {
+	} else { 
 		estimateCoverage = false;
-		if (coverageCutoff < 0 && estimateCutoff)
+		if (coverageCutoff < 0 && estimateCutoff) 
 			coverageCutoff = estimated_cov(graph, directory) / 2;
-		else
+		else 
 			estimateCutoff = false;
 	}
 
@@ -593,7 +580,7 @@ int velvetg(int argc, char **argv)
 	if (minContigLength < 2 * getWordLength(graph))
 		minContigKmerLength = getWordLength(graph);
 	else
-		minContigKmerLength = minContigLength - getWordLength(graph) + 1;
+		minContigKmerLength = minContigLength - getWordLength(graph) + 1;		
 
 	dubious =
 	    removeLowCoverageNodesAndDenounceDubiousReads(graph,
@@ -612,8 +599,7 @@ int velvetg(int argc, char **argv)
 							  lowCovContigsFilename);
 
 	removeHighCoverageNodes(graph, maxCoverageCutoff, exportFilteredNodes, minContigKmerLength, highCovContigsFilename);
-	if (doClipTips)
-		clipTipsHard(graph, conserveLong);
+	clipTipsHard(graph, conserveLong);
 
 	if (sequences->readCount > 0 && sequences->categories[0] == REFERENCE)
 		removeLowArcs(graph, coverageCutoff);
@@ -649,7 +635,7 @@ int velvetg(int argc, char **argv)
 	strcpy(graphFilename, directory);
 	strcat(graphFilename, "/contigs.fa");
 	sequenceLengths = getSequenceLengths(sequences, getWordLength(graph));
-	exportLongNodeSequences(graphFilename, graph, minContigKmerLength, sequences, sequenceLengths, coverageMask);
+	exportLongNodeSequences(graphFilename, graph, minContigKmerLength, sequences, sequenceLengths, coverageMask); 
 
 	if (exportAlignments) {
 		strcpy(graphFilename, directory);
@@ -677,9 +663,9 @@ int velvetg(int argc, char **argv)
 	if (unusedReads)
 		exportUnusedReads(graph, sequences, minContigKmerLength, directory);
 
-	if (estimateCoverage)
+	if (estimateCoverage) 
 		velvetLog("Estimated Coverage = %f\n", expectedCoverage);
-	if (estimateCutoff)
+	if (estimateCutoff) 
 		velvetLog("Estimated Coverage cutoff = %f\n", coverageCutoff);
 
 	logFinalStats(graph, minContigKmerLength, directory);
@@ -687,25 +673,25 @@ int velvetg(int argc, char **argv)
 	if (clean > 0) {
 		strcpy(graphFilename, directory);
 		strcat(graphFilename, "/Roadmaps");
-		remove(graphFilename);
+		remove(graphFilename);	
 
 		strcpy(graphFilename, directory);
 		strcat(graphFilename, "/LastGraph");
-		remove(graphFilename);
-	}
+		remove(graphFilename);	
+	} 
 
 	if (clean > 1) {
 		strcpy(graphFilename, directory);
 		strcat(graphFilename, "/Sequences");
-		remove(graphFilename);
+		remove(graphFilename);	
 
 		strcpy(graphFilename, directory);
 		strcat(graphFilename, "/Graph2");
-		remove(graphFilename);
+		remove(graphFilename);	
 
 		strcpy(graphFilename, directory);
 		strcat(graphFilename, "/Graph");
-		remove(graphFilename);
+		remove(graphFilename);	
 	}
 
 	free(sequenceLengths);

data/ext/src/src/scaffold.h

@@ -18,7 +18,7 @@ Copyright 2007, 2008 Daniel Zerbino (zerbino@ebi.ac.uk)
     Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
 
 */
-#ifndef _SSCAFFOLD_H_
+#ifndef _SCAFFOLD_H_
 #define _SCAFFOLD_H_
 
 typedef struct connection_st Connection;

data/ext/src/src/utility.c

@@ -1,5 +1,5 @@
 /*
-Copyright 2009 John Marshall (jm18@sanger.ac.uk)
+Copyright 2009 John Marshall (jm18@sanger.ac.uk) 
 
     This file is part of Velvet.
 
@@ -89,15 +89,14 @@ void exitErrorf(int exitStatus, boolean showErrno, const char *format, ...)
        fprintf(stderr, "\n");
        va_end(args);
 
-#ifdef DEBUG
+#ifdef DEBUG 
 	abort();
-#endif
+#endif 
        exit(exitStatus);
 }
 
 void velvetLog(const char *format, ...)
 {
-  /*     Don't print anything as it interferes with code bound through bioruby-velvet_underground
   static boolean timeIsSet = false;
   static struct timeval tvStart;
   struct timeval tvNow;
@@ -118,14 +117,14 @@ void velvetLog(const char *format, ...)
   vprintf(format, args);
   va_end(args);
 
-#ifdef DEBUG
+#ifdef DEBUG 
   fflush(stdout);
-#endif*/
+#endif
 }
 
-void velvetFprintf(FILE * file, const char * format, ...)
+void velvetFprintf(FILE * file, const char * format, ...) 
 {
-  va_list args;
+	va_list args;
 
 	va_start(args, format);
 	if (vfprintf(file, format, args) < 0) {
@@ -133,9 +132,9 @@ void velvetFprintf(FILE * file, const char * format, ...)
 		       fprintf(stderr, "%s: ", programName);
 		fprintf(stderr, "Could not write into file\n");
 		va_end(args);
-#ifdef DEBUG
+#ifdef DEBUG 
 		abort();
-#endif
+#endif 
 		exit(EXIT_FAILURE);
 	}
 	va_end(args);

data/lib/bio-velvet_underground/binary_sequence_store.rb

@@ -24,6 +24,18 @@ class Bio::Velvet::Underground
       @readset[:readCount]
     end
 
+    # Return true if paired, else false
+    def paired?(sequence_id)
+      cat = FFI::Pointer.new(:int8, @readset[:categories])[sequence_id-1].read_int8
+      if cat == 0
+        return false
+      elsif cat == 1
+        return true
+      else
+        raise "Unexpected velvet sequence category found: #{cat}"
+      end
+    end
+
     # Returns true if the sequence ID refers to the
     # second in a pair of sequences.
     def is_second_in_pair?(sequence_id)
@@ -33,8 +45,10 @@ class Bio::Velvet::Underground
       Bio::Velvet::Underground.isSecondInPair @readset, sequence_id-1
     end
 
-    # Returns the ID of the given sequence_id's pair
+    # Returns the ID of the given sequence_id's pair, or nil if it is not a
+    # paired sequence
     def pair_id(sequence_id)
+      return nil unless paired?(sequence_id)
       if is_second_in_pair?(sequence_id)
         sequence_id-1
       else
@@ -80,6 +94,9 @@ class Bio::Velvet::Underground
     #			   IDnum	 position);
     attach_function :getTightStringInArray, [:pointer, :int32], :pointer
 
+    # int pairedCategories(ReadSet * reads);
+    attach_function :pairedCategories, [:pointer], :int
+
     # boolean isSecondInPair(ReadSet * reads, IDnum index);
     attach_function :isSecondInPair, [:pointer, :int32], :bool
   end

data/lib/bio-velvet_underground/graph.rb

@@ -119,14 +119,16 @@ class Bio::Velvet::Underground
       end
 
       def fwd_short_reads
+        return @short_reads unless @short_reads.nil?
         array_start_pointer = Bio::Velvet::Underground.getNodeReads @internal_node_struct, @graph.internal_graph_struct
         num_short_reads = Bio::Velvet::Underground.getNodeReadCount @internal_node_struct, @graph.internal_graph_struct
-        short_reads = (0...num_short_reads).collect do |i|
+        struct_size = Bio::Velvet::Underground::ShortReadMarker.size #calculate once for performance
+        @short_reads = 0.step(num_short_reads-1, 1).collect do |i|
           # Use the fact that FFI pointers can do pointer arithmetic
-          pointer = array_start_pointer+(i*Bio::Velvet::Underground::ShortReadMarker.size)
+          pointer = array_start_pointer+(i*struct_size)
           NodedRead.new Bio::Velvet::Underground::ShortReadMarker.new(pointer), true
         end
-        return short_reads
+        return @short_reads
       end
 
       def rev_short_reads
@@ -141,7 +143,6 @@ class Bio::Velvet::Underground
         end
         return reads
       end
-
     end
 
     # TODO: this class is currently unimplemented.