r132: optionally copy FASTA/Q comment to SAM

Author: lh3

bwtsw2.h

@@ -12,7 +12,7 @@
 #define BSW2_FLAG_RESCUED 0x800
 
 typedef struct {
-	int skip_sw:16, hard_clip:16;
+	int skip_sw:8, cpy_cmt:8, hard_clip:16;
 	int a, b, q, r, t, qr, bw, max_ins, max_chain_gap;
 	int z, is, t_seeds, multi_2nd;
 	float mask_level, coef;
@@ -45,7 +45,7 @@ typedef struct {
 
 typedef struct {
 	int l, tid;
-	char *name, *seq, *qual, *sam;
+	char *name, *seq, *qual, *sam, *comment;
 } bsw2seq1_t;
 
 #ifdef __cplusplus

bwtsw2_aux.c

@@ -55,6 +55,7 @@ bsw2opt_t *bsw2_init_opt()
 	o->mask_level = 0.50f; o->coef = 5.5f;
 	o->qr = o->q + o->r; o->n_threads = 1; o->chunk_size = 10000000;
 	o->max_chain_gap = 10000;
+	o->cpy_cmt = 0;
 	return o;
 }
 
@@ -551,14 +552,20 @@ static void print_hits(const bntseq_t *bns, const bsw2opt_t *opt, bsw2seq1_t *ks
 				if (p->flag&0x10) kputc(ks->qual[ks->l - 1 - j], &str);
 				else kputc(ks->qual[j], &str);
 			}
-		} else ksprintf(&str, "\t*");
+		} else kputs("\t*", &str);
 		// print optional tags
 		ksprintf(&str, "\tAS:i:%d\tXS:i:%d\tXF:i:%d\tXE:i:%d\tNM:i:%d", p->G, p->G2, p->flag>>16, p->n_seeds, q->nm);
 		if (q->nn) ksprintf(&str, "\tXN:i:%d", q->nn);
 		if (p->l) ksprintf(&str, "\tXI:i:%d", p->l - p->k + 1);
 		if (p->flag&BSW2_FLAG_MATESW) type |= 1;
 		if (p->flag&BSW2_FLAG_TANDEM) type |= 2;
 		if (type) ksprintf(&str, "\tXT:i:%d", type);
+		if (opt->cpy_cmt && ks->comment) {
+			int l = strlen(ks->comment);
+			if (l >= 6 && ks->comment[2] == ':' && ks->comment[4] == ':') {
+				kputc('\t', &str); kputs(ks->comment, &str);
+			}
+		}
 		kputc('\n', &str);
 	}
 	ks->sam = str.s;
@@ -756,6 +763,7 @@ static void kseq_to_bsw2seq(const kseq_t *ks, bsw2seq1_t *p)
 	p->name = strdup(ks->name.s);
 	p->seq = strdup(ks->seq.s);
 	p->qual = ks->qual.l? strdup(ks->qual.s) : 0;
+	p->comment = ks->comment.l? strdup(ks->comment.s) : 0;
 	p->sam = 0;
 }
 

bwtsw2_main.c

@@ -18,7 +18,7 @@ int bwa_bwtsw2(int argc, char *argv[])
 
 	opt = bsw2_init_opt();
 	srand48(11);
-	while ((c = getopt(argc, argv, "q:r:a:b:t:T:w:d:z:m:s:c:N:Hf:MI:SG:")) >= 0) {
+	while ((c = getopt(argc, argv, "q:r:a:b:t:T:w:d:z:m:s:c:N:Hf:MI:SG:C")) >= 0) {
 		switch (c) {
 		case 'q': opt->q = atoi(optarg); break;
 		case 'r': opt->r = atoi(optarg); break;
@@ -37,6 +37,7 @@ int bwa_bwtsw2(int argc, char *argv[])
 		case 'f': xreopen(optarg, "w", stdout); break;
 		case 'I': opt->max_ins = atoi(optarg); break;
 		case 'S': opt->skip_sw = 1; break;
+		case 'C': opt->cpy_cmt = 1; break;
 		case 'G': opt->max_chain_gap = atoi(optarg); break;
 		}
 	}
@@ -55,6 +56,7 @@ int bwa_bwtsw2(int argc, char *argv[])
 		fprintf(stderr, "         -t INT   number of threads [%d]\n", opt->n_threads);
 		fprintf(stderr, "         -f FILE  file to output results to instead of stdout\n");
 		fprintf(stderr, "         -H       in SAM output, use hard clipping instead of soft clipping\n");
+		fprintf(stderr, "         -C       copy FASTA/Q comment to SAM output\n");
 		fprintf(stderr, "         -M       mark multi-part alignments as secondary\n");
 		fprintf(stderr, "         -S       skip Smith-Waterman read pairing\n");
 		fprintf(stderr, "         -I INT   ignore pairs with insert >=INT for inferring the size distr [%d]\n", opt->max_ins);

main.c

@@ -4,7 +4,7 @@
 #include "utils.h"
 
 #ifndef PACKAGE_VERSION
-#define PACKAGE_VERSION "0.6.2-r131"
+#define PACKAGE_VERSION "0.6.2-r132"
 #endif
 
 static int usage()