Add support for interleaved fastq

ebwt_search.cpp

@@ -338,7 +338,8 @@ enum {
 	ARG_QUALS2,
 	ARG_ALLOW_CONTAIN,
 	ARG_COLOR_PRIMER,
-	ARG_WRAPPER
+	ARG_WRAPPER,
+	ARG_INTERLEAVED_FASTQ,
 };
 
 static struct option long_options[] = {
@@ -444,6 +445,7 @@ static struct option long_options[] = {
 	{(char*)"allow-contain",no_argument,       0,            ARG_ALLOW_CONTAIN},
 	{(char*)"col-primer",   no_argument,       0,            ARG_COLOR_PRIMER},
 	{(char*)"wrapper",      required_argument, 0,            ARG_WRAPPER},
+	{(char*)"interleaved",  required_argument, 0,            ARG_INTERLEAVED_FASTQ},
 	{(char*)0, 0, 0, 0} // terminator
 };
 
@@ -461,14 +463,15 @@ static void printUsage(ostream& out) {
 	}
 
 	out << "Usage: " << endl
-        << tool_name << " [options]* <ebwt> {-1 <m1> -2 <m2> | --12 <r> | <s>} [<hit>]" << endl
+        << tool_name << " [options]* <ebwt> {-1 <m1> -2 <m2> | --12 <r> | --interleaved <i> | <s>} [<hit>]" << endl
         << endl
 	    << "  <m1>    Comma-separated list of files containing upstream mates (or the" << endl
 	    << "          sequences themselves, if -c is set) paired with mates in <m2>" << endl
 	    << "  <m2>    Comma-separated list of files containing downstream mates (or the" << endl
 	    << "          sequences themselves if -c is set) paired with mates in <m1>" << endl
 	    << "  <r>     Comma-separated list of files containing Crossbow-style reads.  Can be" << endl
 	    << "          a mixture of paired and unpaired.  Specify \"-\" for stdin." << endl
+	    << "  <i>     Files with interleaved paired-end FASTQ reads." << endl
 	    << "  <s>     Comma-separated list of files containing unpaired reads, or the" << endl
 	    << "          sequences themselves, if -c is set.  Specify \"-\" for stdin." << endl
 	    << "  <hit>   File to write hits to (default: stdout)" << endl
@@ -647,6 +650,7 @@ static void parseOptions(int argc, const char **argv) {
 			case '1': tokenize(optarg, ",", mates1); break;
 			case '2': tokenize(optarg, ",", mates2); break;
 			case ARG_ONETWO: tokenize(optarg, ",", mates12); format = TAB_MATE; break;
+			case ARG_INTERLEAVED_FASTQ: tokenize(optarg, ",", mates12); format = INTERLEAVED; break;
 			case 'f': format = FASTA; break;
 			case 'F': {
 				format = FASTA_CONT;
@@ -2592,6 +2596,12 @@ patsrcFromStrings(int format,
 			                               trim3, trim5,
 			                               solexaQuals, phred64Quals,
 			                               integerQuals);
+		case INTERLEAVED:
+			return new FastqPatternSource (reads, color,
+			                               patDumpfile,
+			                               trim3, trim5,
+			                               solexaQuals, phred64Quals,
+			                               integerQuals, true /* is interleaved */);
 		case TAB_MATE:
 			return new TabbedPatternSource(reads, false, color,
 			                               patDumpfile,

formats.h

@@ -12,6 +12,7 @@ enum file_format {
 	FASTA = 1,
 	FASTA_CONT,
 	FASTQ,
+	INTERLEAVED,
 	TAB_MATE,
 	RAW,
 	CMDLINE,

hit.h

@@ -882,6 +882,7 @@ class HitSinkPerThread {
 		hitsForThisRead_(),
 		_max(max),
 		_n(n),
+        defaultMapq_(defaultMapq),
 		threadId_(threadId)
 	{
 		sink.addWrapper();

pat.cpp

@@ -906,7 +906,7 @@ pair<bool, int> FastqPatternSource::nextBatchFromFile(
 	bool batch_a)
 {
 	int c = 0;
-	vector<Read>& readBuf = batch_a ? pt.bufa_ : pt.bufb_;
+	vector<Read>* readBuf = batch_a ? &pt.bufa_ : &pt.bufb_;
 	if(first_) {
 		c = getc_wrapper();
 		while(c == '\r' || c == '\n') {
@@ -917,15 +917,15 @@ pair<bool, int> FastqPatternSource::nextBatchFromFile(
 			throw 1;
 		}
 		first_ = false;
-		readBuf[0].readOrigBuf[0] = c;
-		readBuf[0].readOrigBufLen = 1;
+		(*readBuf)[0].readOrigBuf[0] = c;
+		(*readBuf)[0].readOrigBufLen = 1;
 	}
 	bool done = false, aborted = false;
 	size_t readi = 0;
 	// Read until we run out of input or until we've filled the buffer
-	for(; readi < pt.max_buf_ && !done; readi++) {
-		char* buf = readBuf[readi].readOrigBuf;
-		assert(readi == 0 || readBuf[readi].readOrigBufLen == 0);
+	while (readi < pt.max_buf_ && !done) {
+		char* buf = (*readBuf)[readi].readOrigBuf;
+		assert(readi == 0 || (*readBuf)[readi].readOrigBufLen == 0);
 		int newlines = 4;
 		while(newlines) {
 			c = getc_wrapper();
@@ -936,10 +936,26 @@ pair<bool, int> FastqPatternSource::nextBatchFromFile(
 				newlines--;
 				c = '\n';
 			} else if(done) {
-				aborted = true; // Unexpected EOF
+				if (newlines == 4) {
+					newlines = 0;
+				} else {
+					aborted = true; // Unexpected EOF
+				}
 				break;
 			}
-			buf[readBuf[readi].readOrigBufLen++] = c;
+			buf[(*readBuf)[readi].readOrigBufLen++] = c;
+		}
+		if (c > 0) {
+			if (interleaved_) {
+				// alternate between read buffers
+				batch_a = !batch_a;
+				readBuf = batch_a ? &pt.bufa_ : &pt.bufb_;
+				// increment read counter after each pair gets read
+				readi = batch_a ? readi + 1 : readi;
+			}
+			else {
+				readi++;
+			}
 		}
 	}
 	if(aborted) {

pat.h

@@ -934,6 +934,7 @@ class FastqPatternSource : public CFilePatternSource {
 	    bool solexa_quals = false,
 	    bool phred64Quals = false,
 	    bool integer_quals = false,
+	    bool interleaved = false,
 	    uint32_t skip = 0) :
 		CFilePatternSource(
 			infiles,
@@ -945,6 +946,7 @@ class FastqPatternSource : public CFilePatternSource {
 		solQuals_(solexa_quals),
 		phred64Quals_(phred64Quals),
 		intQuals_(integer_quals),
+		interleaved_(interleaved),
 		color_(color) { }
 	
 	virtual void reset() {
@@ -988,6 +990,7 @@ class FastqPatternSource : public CFilePatternSource {
 	bool solQuals_;
 	bool phred64Quals_;
 	bool intQuals_;
+	bool interleaved_;
 	bool color_;
 };