X-Git-Url: https://git.donarmstrong.com/?a=blobdiff_plain;f=bam_plcmd.c;h=ccaa28dec1dffc5174d77c31dc81603f8214b38c;hb=88abd40bb30422736fc8f1aa1f40ea6ece050bac;hp=46f078a99e715a89ccc4f73119192fe8d1ef280c;hpb=4c4b72301b736e08e37b5e0dd1d435477d969898;p=samtools.git

diff --git a/bam_plcmd.c b/bam_plcmd.c
index 46f078a..ccaa28d 100644
--- a/bam_plcmd.c
+++ b/bam_plcmd.c
@@ -17,6 +17,11 @@ KHASH_MAP_INIT_INT64(64, indel_list_t)
 #define BAM_PLF_INDEL_ONLY 0x04
 #define BAM_PLF_GLF        0x08
 #define BAM_PLF_VAR_ONLY   0x10
+#define BAM_PLF_2ND        0x20
+#define BAM_PLF_RANBASE    0x40
+#define BAM_PLF_1STBASE    0x80
+#define BAM_PLF_ALLBASE    0x100
+#define BAM_PLF_READPOS    0x200
 
 typedef struct {
 	bam_header_t *h;
@@ -27,6 +32,7 @@ typedef struct {
 	uint32_t format;
 	int tid, len, last_pos;
 	int mask;
+    int max_depth;  // for indel calling, ignore reads with the depth too high. 0 for unlimited
 	char *ref;
 	glfFile fp_glf; // for glf output only
 } pu_data_t;
@@ -120,9 +126,12 @@ static int glt3_func(uint32_t tid, uint32_t pos, int n, const bam_pileup1_t *pu,
 	g3->offset = pos - d->last_pos;
 	d->last_pos = pos;
 	glf3_write1(d->fp_glf, g3);
-	if (proposed_indels)
-		r = bam_maqindel(n, pos, d->ido, pu, d->ref, proposed_indels[0], proposed_indels+1);
-	else r = bam_maqindel(n, pos, d->ido, pu, d->ref, 0, 0);
+    if (pos < d->len) {
+        int m = (!d->max_depth || d->max_depth>n) ? n : d->max_depth;
+		if (proposed_indels)
+			r = bam_maqindel(m, pos, d->ido, pu, d->ref, proposed_indels[0], proposed_indels+1);
+		else r = bam_maqindel(m, pos, d->ido, pu, d->ref, 0, 0);
+	}
 	if (r) { // then write indel line
 		int het = 3 * n, min;
 		min = het;
@@ -168,7 +177,7 @@ static int pileup_func(uint32_t tid, uint32_t pos, int n, const bam_pileup1_t *p
 	// update d->ref if necessary
 	if (d->fai && (int)tid != d->tid) {
 		free(d->ref);
-		d->ref = fai_fetch(d->fai, d->h->target_name[tid], &d->len);
+		d->ref = faidx_fetch_seq(d->fai, d->h->target_name[tid], 0, 0x7fffffff, &d->len);
 		d->tid = tid;
 	}
 	rb = (d->ref && (int)pos < d->len)? d->ref[pos] : 'N';
@@ -179,18 +188,39 @@ static int pileup_func(uint32_t tid, uint32_t pos, int n, const bam_pileup1_t *p
 		if (i == n) return 0;
 	}
 	// call the consensus and indel
-	if (d->format & BAM_PLF_CNS) // call consensus
-		cns = bam_maqcns_call(n, pu, d->c);
-	if ((d->format & (BAM_PLF_CNS|BAM_PLF_INDEL_ONLY)) && d->ref) { // call indels
-		if (proposed_indels) // the first element gives the size of the array
-			r = bam_maqindel(n, pos, d->ido, pu, d->ref, proposed_indels[0], proposed_indels+1);
-		else r = bam_maqindel(n, pos, d->ido, pu, d->ref, 0, 0);
+	if (d->format & BAM_PLF_CNS) { // call consensus
+		if (d->format & (BAM_PLF_RANBASE|BAM_PLF_1STBASE)) { // use a random base or the 1st base as the consensus call
+			const bam_pileup1_t *p = (d->format & BAM_PLF_1STBASE)? pu : pu + (int)(drand48() * n);
+			int q = bam1_qual(p->b)[p->qpos];
+			int mapQ = p->b->core.qual < d->c->cap_mapQ? p->b->core.qual : d->c->cap_mapQ;
+			uint32_t b = bam1_seqi(bam1_seq(p->b), p->qpos);
+			cns = b<<28 | 0xf<<24 | mapQ<<16 | q<<8;
+		} else if (d->format & BAM_PLF_ALLBASE) { // collapse all bases
+			uint64_t rmsQ = 0;
+			uint32_t b = 0;
+			for (i = 0; i < n; ++i) {
+				const bam_pileup1_t *p = pu + i;
+				int q = p->b->core.qual < d->c->cap_mapQ? p->b->core.qual : d->c->cap_mapQ;
+				b |= bam1_seqi(bam1_seq(p->b), p->qpos);
+				rmsQ += q * q;
+			}
+			rmsQ = (uint64_t)(sqrt((double)rmsQ / n) + .499);
+			cns = b<<28 | 0xf<<24 | rmsQ<<16 | 60<<8;
+		} else cns = bam_maqcns_call(n, pu, d->c);
+	}
+    if ((d->format & (BAM_PLF_CNS|BAM_PLF_INDEL_ONLY)) && d->ref && pos < d->len) { // call indels
+        int m = (!d->max_depth || d->max_depth>n) ? n : d->max_depth;
+        if (proposed_indels) // the first element gives the size of the array
+            r = bam_maqindel(m, pos, d->ido, pu, d->ref, proposed_indels[0], proposed_indels+1);
+        else r = bam_maqindel(m, pos, d->ido, pu, d->ref, 0, 0);
 	}
 	// when only variant sites are asked for, test if the site is a variant
 	if ((d->format & BAM_PLF_CNS) && (d->format & BAM_PLF_VAR_ONLY)) {
 		if (!(bam_nt16_table[rb] != 15 && cns>>28 != bam_nt16_table[rb])) { // not a SNP
-			if (!(r && (r->gt == 2 || strcmp(r->s[r->gt], "*")))) // not an indel
+			if (!(r && (r->gt == 2 || strcmp(r->s[r->gt], "*")))) { // not an indel
+				if (r) bam_maqindel_ret_destroy(r);
 				return 0;
+			}
 		}
 	}
 	// print the first 3 columns
@@ -246,6 +276,21 @@ static int pileup_func(uint32_t tid, uint32_t pos, int n, const bam_pileup1_t *p
 		if (c > 126) c = 126;
 		putchar(c);
 	}
+	if (d->format & BAM_PLF_2ND) { // print 2nd calls and qualities
+		const unsigned char *q;
+		putchar('\t');
+		for (i = 0; i < n; ++i) {
+			const bam_pileup1_t *p = pu + i;
+			q = bam_aux_get(p->b, "E2");
+			putchar(q? q[p->qpos + 1] : 'N');
+		}
+		putchar('\t');
+		for (i = 0; i < n; ++i) {
+			const bam_pileup1_t *p = pu + i;
+			q = bam_aux_get(p->b, "U2");
+			putchar(q? q[p->qpos + 1] : '!');
+		}
+	}
 	// print mapping quality if -s is flagged on the command line
 	if (d->format & BAM_PLF_SIMPLE) {
 		putchar('\t');
@@ -255,6 +300,15 @@ static int pileup_func(uint32_t tid, uint32_t pos, int n, const bam_pileup1_t *p
 			putchar(c);
 		}
 	}
+	// print read position
+	if (d->format & BAM_PLF_READPOS) {
+		putchar('\t');
+		for (i = 0; i < n; ++i) {
+			int x = pu[i].qpos;
+			int l = pu[i].b->core.l_qseq;
+			printf("%d,", x < l/2? x+1 : -((l-1)-x+1));
+		}
+	}
 	putchar('\n');
 	// print the indel line if r has been calculated. This only happens if:
 	// a) -c or -i are flagged, AND b) the reference sequence is available
@@ -266,7 +320,8 @@ static int pileup_func(uint32_t tid, uint32_t pos, int n, const bam_pileup1_t *p
 		printf("%d\t%d\t", rms_mapq, n);
 		printf("%s\t%s\t", r->s[0], r->s[1]);
 		//printf("%d\t%d\t", r->gl[0], r->gl[1]);
-		printf("%d\t%d\t%d\n", r->cnt1, r->cnt2, r->cnt_anti);
+		printf("%d\t%d\t%d\t", r->cnt1, r->cnt2, r->cnt_anti);
+		printf("%d\t%d\n", r->cnt_ref, r->cnt_ambi);
 		bam_maqindel_ret_destroy(r);
 	}
 	return 0;
@@ -274,14 +329,16 @@ static int pileup_func(uint32_t tid, uint32_t pos, int n, const bam_pileup1_t *p
 
 int bam_pileup(int argc, char *argv[])
 {
-	int c;
+	int c, is_SAM = 0;
 	char *fn_list = 0, *fn_fa = 0, *fn_pos = 0;
 	pu_data_t *d = (pu_data_t*)calloc(1, sizeof(pu_data_t));
+    d->max_depth = 0;
 	d->tid = -1; d->mask = BAM_DEF_MASK;
 	d->c = bam_maqcns_init();
 	d->ido = bam_maqindel_opt_init();
-	while ((c = getopt(argc, argv, "st:f:cT:N:r:l:im:gI:G:vM:")) >= 0) {
+	while ((c = getopt(argc, argv, "st:f:cT:N:r:l:d:im:gI:G:vM:S2aR:P")) >= 0) {
 		switch (c) {
+		case 'a': d->c->is_soap = 1; break;
 		case 's': d->format |= BAM_PLF_SIMPLE; break;
 		case 't': fn_list = strdup(optarg); break;
 		case 'l': fn_pos = strdup(optarg); break;
@@ -290,24 +347,39 @@ int bam_pileup(int argc, char *argv[])
 		case 'N': d->c->n_hap = atoi(optarg); break;
 		case 'r': d->c->het_rate = atof(optarg); break;
 		case 'M': d->c->cap_mapQ = atoi(optarg); break;
+		case 'd': d->max_depth = atoi(optarg); break;
 		case 'c': d->format |= BAM_PLF_CNS; break;
 		case 'i': d->format |= BAM_PLF_INDEL_ONLY; break;
 		case 'v': d->format |= BAM_PLF_VAR_ONLY; break;
 		case 'm': d->mask = strtol(optarg, 0, 0); break;
 		case 'g': d->format |= BAM_PLF_GLF; break;
+		case '2': d->format |= BAM_PLF_2ND; break;
+		case 'P': d->format |= BAM_PLF_READPOS; break;
 		case 'I': d->ido->q_indel = atoi(optarg); break;
 		case 'G': d->ido->r_indel = atof(optarg); break;
+		case 'S': is_SAM = 1; break;
+		case 'R':
+			if (strcmp(optarg, "random") == 0) d->format |= BAM_PLF_RANBASE;
+			else if (strcmp(optarg, "first") == 0) d->format |= BAM_PLF_1STBASE;
+			else if (strcmp(optarg, "all") == 0) d->format |= BAM_PLF_ALLBASE;
+			else fprintf(stderr, "[bam_pileup] unrecognized -R\n");
+			break;
 		default: fprintf(stderr, "Unrecognizd option '-%c'.\n", c); return 1;
 		}
 	}
+	if (fn_list) is_SAM = 1;
 	if (optind == argc) {
 		fprintf(stderr, "\n");
 		fprintf(stderr, "Usage:  samtools pileup [options] <in.bam>|<in.sam>\n\n");
 		fprintf(stderr, "Option: -s        simple (yet incomplete) pileup format\n");
+		fprintf(stderr, "        -S        the input is in SAM\n");
+		fprintf(stderr, "        -a        use the SOAPsnp model for SNP calling\n");
+		fprintf(stderr, "        -2        output the 2nd best call and quality\n");
 		fprintf(stderr, "        -i        only show lines/consensus with indels\n");
 		fprintf(stderr, "        -m INT    filtering reads with bits in INT [%d]\n", d->mask);
 		fprintf(stderr, "        -M INT    cap mapping quality at INT [%d]\n", d->c->cap_mapQ);
-		fprintf(stderr, "        -t FILE   list of reference sequences (assume the input is in SAM)\n");
+        fprintf(stderr, "        -d INT    limit maximum depth for indels [unlimited]\n");
+		fprintf(stderr, "        -t FILE   list of reference sequences (force -S)\n");
 		fprintf(stderr, "        -l FILE   list of sites at which pileup is output\n");
 		fprintf(stderr, "        -f FILE   reference sequence in the FASTA format\n\n");
 		fprintf(stderr, "        -c        output the maq consensus sequence\n");
@@ -322,6 +394,7 @@ int bam_pileup(int argc, char *argv[])
 		free(fn_list); free(fn_fa); free(d);
 		return 1;
 	}
+	if (d->format & (BAM_PLF_RANBASE|BAM_PLF_1STBASE|BAM_PLF_ALLBASE)) d->format |= BAM_PLF_CNS;
 	if (fn_fa) d->fai = fai_load(fn_fa);
 	if (d->format & (BAM_PLF_CNS|BAM_PLF_GLF)) bam_maqcns_prepare(d->c); // consensus calling
 	if (d->format & BAM_PLF_GLF) { // for glf output
@@ -333,9 +406,11 @@ int bam_pileup(int argc, char *argv[])
 	}
 	if (d->fai == 0 && (d->format & (BAM_PLF_CNS|BAM_PLF_INDEL_ONLY)))
 		fprintf(stderr, "[bam_pileup] indels will not be called when -f is absent.\n");
+	if (fn_fa && is_SAM && fn_list == 0) fn_list = samfaipath(fn_fa);
+
 	{
 		samfile_t *fp;
-		fp = fn_list? samopen(argv[optind], "r", fn_list) : samopen(argv[optind], "rb", 0);
+		fp = is_SAM? samopen(argv[optind], "r", fn_list) : samopen(argv[optind], "rb", 0);
 		if (fp == 0 || fp->header == 0) {
 			fprintf(stderr, "[bam_pileup] fail to read the header: non-exisiting file or wrong format.\n");
 			return 1;