Merge branch 'master' into master_fixes

bwa.1

@@ -1,4 +1,4 @@
-.TH bwa 1 "15 March 2013" "bwa-0.7.3a" "Bioinformatics tools"
+.TH bwa 1 "15 March 2013" "bwa-0.7.4" "Bioinformatics tools"
 .SH NAME
 .PP
 bwa - Burrows-Wheeler Alignment Tool
@@ -220,7 +220,7 @@ deducted. [5]
 Penalty for an unpaired read pair. BWA-MEM scores an unpaired read pair as
 .RI scoreRead1+scoreRead2- INT
 and scores a paired as scoreRead1+scoreRead2-insertPenalty. It compares these
-two scores to determine whether we should force pairing. [9]
+two scores to determine whether we should force pairing. [17]
 .TP
 .B -p
 Assume the first input query file is interleaved paired-end FASTA/Q. See the command description for details.

bwamem.c

@@ -46,7 +46,7 @@ mem_opt_t *mem_opt_init()
 	o->a = 1; o->b = 4; o->q = 6; o->r = 1; o->w = 100;
 	o->T = 30;
 	o->zdrop = 100;
-	o->pen_unpaired = 9;
+	o->pen_unpaired = 17;
 	o->pen_clip = 5;
 	o->min_seed_len = 19;
 	o->split_width = 10;

bwape.c

@@ -106,6 +106,11 @@ static int infer_isize(int n_seqs, bwa_seq_t *seqs[2], isize_info_t *ii, double
 	tmp  = (int)(p25 - OUTLIER_BOUND * (p75 - p25) + .499);
 	ii->low = tmp > max_len? tmp : max_len; // ii->low is unsigned
 	ii->high = (int)(p75 + OUTLIER_BOUND * (p75 - p25) + .499);
+	if (ii->low > ii->high) {
+		fprintf(stderr, "[infer_isize] fail to infer insert size: upper bound is smaller than read length\n");
+		free(isizes);
+		return -1;
+	}
 	for (i = 0, x = n = 0; i < tot; ++i)
 		if (isizes[i] >= ii->low && isizes[i] <= ii->high)
 			++n, x += isizes[i];
@@ -404,7 +409,7 @@ bwa_cigar_t *bwa_sw_core(bwtint_t l_pac, const ubyte_t *pacseq, int len, const u
 	bwa_cigar_t *cigar = 0;
 	ubyte_t *ref_seq;
 	bwtint_t k, x, y, l;
-	int xtra;
+	int xtra, gscore;
 	int8_t mat[25];
 
 	bwa_fill_scmat(1, 3, mat);
@@ -422,12 +427,12 @@ bwa_cigar_t *bwa_sw_core(bwtint_t l_pac, const ubyte_t *pacseq, int len, const u
 	// do alignment
 	xtra = KSW_XSUBO | KSW_XSTART | (len < 250? KSW_XBYTE : 0);
 	r = ksw_align(len, (uint8_t*)seq, l, ref_seq, 5, mat, 5, 1, xtra, 0);
-	ksw_global(r.qe - r.qb + 1, &seq[r.qb], r.te - r.tb + 1, &ref_seq[r.tb], 5, mat, 5, 1, 50, n_cigar, &cigar32);
+	gscore = ksw_global(r.qe - r.qb + 1, &seq[r.qb], r.te - r.tb + 1, &ref_seq[r.tb], 5, mat, 5, 1, 50, n_cigar, &cigar32);
 	cigar = (bwa_cigar_t*)cigar32;
 	for (k = 0; k < *n_cigar; ++k)
 		cigar[k] = __cigar_create((cigar32[k]&0xf), (cigar32[k]>>4));
 
-	if (r.score < SW_MIN_MATCH_LEN || r.score2 == r.score) { // poor hit or tandem hits
+	if (r.score < SW_MIN_MATCH_LEN || r.score2 == r.score || gscore != r.score) { // poor hit or tandem hits or weird alignment
 		free(cigar); free(ref_seq); *n_cigar = 0;
 		return 0;
 	}

bwtsw2_pair.c

@@ -47,7 +47,7 @@ bsw2pestat_t bsw2_stat(int n, bwtsw2_t **buf, kstring_t *msg, int max_ins)
 	p75 = isize[(int)(.75 * k + .499)];
 	ksprintf(msg, "[%s] infer the insert size distribution from %d high-quality pairs.\n", __func__, k);
 	if (k < 8) {
-		ksprintf(msg, "[%s] fail to infer the insert size distribution.\n", __func__);
+		ksprintf(msg, "[%s] fail to infer the insert size distribution: too few good pairs.\n", __func__);
 		free(isize);
 		r.failed = 1;
 		return r;
@@ -56,6 +56,12 @@ bsw2pestat_t bsw2_stat(int n, bwtsw2_t **buf, kstring_t *msg, int max_ins)
 	r.low  = tmp > max_len? tmp : max_len;
 	if (r.low < 1) r.low = 1;
 	r.high = (int)(p75 + OUTLIER_BOUND * (p75 - p25) + .499);
+	if (r.low > r.high) {
+		ksprintf(msg, "[%s] fail to infer the insert size distribution: upper bound is smaller than max read length.\n", __func__);
+		free(isize);
+		r.failed = 1;
+		return r;
+	}
 	ksprintf(msg, "[%s] (25, 50, 75) percentile: (%d, %d, %d)\n", __func__, p25, p50, p75);
 	ksprintf(msg, "[%s] low and high boundaries for computing mean and std.dev: (%d, %d)\n", __func__, r.low, r.high);
 	for (i = x = 0, r.avg = 0; i < k; ++i)

ksw.c

@@ -202,10 +202,11 @@ kswr_t ksw_u8(kswq_t *q, int tlen, const uint8_t *target, int _gapo, int _gape,
 	r.score = gmax + q->shift < 255? gmax : 255;
 	r.te = te;
 	if (r.score != 255) { // get a->qe, the end of query match; find the 2nd best score
-		int max = -1, low, high, qlen = slen * 16;
+		int max = -1, tmp, low, high, qlen = slen * 16;
 		uint8_t *t = (uint8_t*)Hmax;
 		for (i = 0; i < qlen; ++i, ++t)
 			if ((int)*t > max) max = *t, r.qe = i / 16 + i % 16 * slen;
+			else if ((int)*t == max && (tmp = i / 16 + i % 16 * slen) < r.qe) r.qe = tmp; 
 		//printf("%d,%d\n", max, gmax);
 		if (b) {
 			i = (r.score + q->max - 1) / q->max;
@@ -303,10 +304,11 @@ kswr_t ksw_i16(kswq_t *q, int tlen, const uint8_t *target, int _gapo, int _gape,
 	}
 	r.score = gmax; r.te = te;
 	{
-		int max = -1, low, high, qlen = slen * 8;
+		int max = -1, tmp, low, high, qlen = slen * 8;
 		uint16_t *t = (uint16_t*)Hmax;
 		for (i = 0, r.qe = -1; i < qlen; ++i, ++t)
 			if ((int)*t > max) max = *t, r.qe = i / 8 + i % 8 * slen;
+			else if ((int)*t == max && (tmp = i / 8 + i % 8 * slen) < r.qe) r.qe = tmp; 
 		if (b) {
 			i = (r.score + q->max - 1) / q->max;
 			low = te - i; high = te + i;

main.c

@@ -3,7 +3,7 @@
 #include "utils.h"
 
 #ifndef PACKAGE_VERSION
-#define PACKAGE_VERSION "0.7.3-r376-beta"
+#define PACKAGE_VERSION "0.7.3-r382-beta"
 #endif
 
 int bwa_fa2pac(int argc, char *argv[]);