X-Git-Url: https://git.donarmstrong.com/?a=blobdiff_plain;f=bam_plcmd.c;h=80799e3b28af95208796424c31a4bf5b7dd24ba0;hb=63af098c78ef4163356c281e87bfb467adc8824d;hp=99592276c0bded2906122a09528f5453bdb7ebd3;hpb=754152319e7a5f1e48086b6508cd3d2f7e839349;p=samtools.git

diff --git a/bam_plcmd.c b/bam_plcmd.c
index 9959227..80799e3 100644
--- a/bam_plcmd.c
+++ b/bam_plcmd.c
@@ -5,6 +5,7 @@
 #include "sam.h"
 #include "faidx.h"
 #include "bam_maqcns.h"
+#include "bam2bcf.h"
 #include "khash.h"
 #include "glf.h"
 #include "kstring.h"
@@ -341,12 +342,12 @@ int bam_pileup(int argc, char *argv[])
     d->max_depth = 0;
 	d->tid = -1; d->mask = BAM_DEF_MASK;
 	d->c = bam_maqcns_init();
-	d->c->is_soap = 1; // change the default model
+	d->c->errmod = BAM_ERRMOD_MAQ2; // change the default model
 	d->ido = bam_maqindel_opt_init();
 	while ((c = getopt(argc, argv, "st:f:cT:N:r:l:d:im:gI:G:vM:S2aR:PA")) >= 0) {
 		switch (c) {
-		case 'a': d->c->is_soap = 1; break;
-		case 'A': d->c->is_soap = 0; break;
+		case 'a': d->c->errmod = BAM_ERRMOD_SOAP; break;
+		case 'A': d->c->errmod = BAM_ERRMOD_MAQ; break;
 		case 's': d->format |= BAM_PLF_SIMPLE; break;
 		case 't': fn_list = strdup(optarg); break;
 		case 'l': fn_pos = strdup(optarg); break;
@@ -390,7 +391,7 @@ int bam_pileup(int argc, char *argv[])
 		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");
+		fprintf(stderr, "        -c        output the SOAPsnp consensus sequence\n");
 		fprintf(stderr, "        -v        print variants only (for -c)\n");
 		fprintf(stderr, "        -g        output in the GLFv3 format (suppressing -c/-i/-s)\n");
 		fprintf(stderr, "        -T FLOAT  theta in maq consensus calling model (for -c/-g) [%f]\n", d->c->theta);
@@ -448,21 +449,31 @@ int bam_pileup(int argc, char *argv[])
  * mpileup *
  ***********/
 
+#define MPLP_GLF   0x10
+#define MPLP_NO_COMP 0x20
+
 typedef struct {
-	char *reg;
+	int max_mq, min_mq, flag, min_baseQ;
+	double theta;
+	char *reg, *fn_pos;
 	faidx_t *fai;
+	kh_64_t *hash;
 } mplp_conf_t;
 
 typedef struct {
 	bamFile fp;
 	bam_iter_t iter;
+	int min_mq;
 } mplp_aux_t;
 
 static int mplp_func(void *data, bam1_t *b)
 {
 	mplp_aux_t *ma = (mplp_aux_t*)data;
-	if (ma->iter) return bam_iter_read(ma->fp, ma->iter, b);
-	return bam_read1(ma->fp, b);
+	int ret;
+	do {
+		ret = ma->iter? bam_iter_read(ma->fp, ma->iter, b) : bam_read1(ma->fp, b);
+	} while (b->core.qual < ma->min_mq && ret >= 0);
+	return ret;
 }
 
 static int mpileup(mplp_conf_t *conf, int n, char **fn)
@@ -473,14 +484,24 @@ static int mpileup(mplp_conf_t *conf, int n, char **fn)
 	bam_mplp_t iter;
 	bam_header_t *h = 0;
 	char *ref;
-	// allocate
+	khash_t(64) *hash = 0;
+
+	bcf_callaux_t *bca = 0;
+	bcf_callret1_t *bcr = 0;
+	bcf_call_t bc;
+	bcf_t *bp = 0;
+	bcf_hdr_t *bh = 0;
+
+	memset(&bc, 0, sizeof(bcf_call_t));
 	data = calloc(n, sizeof(void*));
 	plp = calloc(n, sizeof(void*));
 	n_plp = calloc(n, sizeof(int*));
+
 	// read the header and initialize data
 	for (i = 0; i < n; ++i) {
 		bam_header_t *h_tmp;
 		data[i] = calloc(1, sizeof(mplp_aux_t));
+		data[i]->min_mq = conf->min_mq;
 		data[i]->fp = bam_open(fn[i], "r");
 		h_tmp = bam_header_read(data[i]->fp);
 		if (conf->reg) {
@@ -505,35 +526,96 @@ static int mpileup(mplp_conf_t *conf, int n, char **fn)
 			bam_header_destroy(h_tmp);
 		}
 	}
+	if (conf->fn_pos) hash = load_pos(conf->fn_pos, h);
+	// write the VCF header
+	if (conf->flag & MPLP_GLF) {
+		kstring_t s;
+		bh = calloc(1, sizeof(bcf_hdr_t));
+		s.l = s.m = 0; s.s = 0;
+		bp = bcf_open("-", (conf->flag&MPLP_NO_COMP)? "wu" : "w");
+		for (i = 0; i < h->n_targets; ++i) {
+			kputs(h->target_name[i], &s);
+			kputc('\0', &s);
+		}
+		bh->l_nm = s.l;
+		bh->name = malloc(s.l);
+		memcpy(bh->name, s.s, s.l);
+		s.l = 0;
+		for (i = 0; i < n; ++i) {
+			const char *p;
+			if ((p = strstr(fn[i], ".bam")) != 0)
+				kputsn(fn[i], p - fn[i], &s);
+			else kputs(fn[i], &s);
+			kputc('\0', &s);
+		}
+		bh->l_smpl = s.l;
+		bh->sname = malloc(s.l);
+		memcpy(bh->sname, s.s, s.l);
+		bh->l_txt = 0;
+		free(s.s);
+		bcf_hdr_sync(bh);
+		bcf_hdr_write(bp, bh);
+	}
 	// mpileup
+	if (conf->flag & MPLP_GLF) {
+		bca = bcf_call_init(-1., conf->min_baseQ);
+		bcr = calloc(n, sizeof(bcf_callret1_t));
+	}
 	ref_tid = -1; ref = 0;
 	iter = bam_mplp_init(n, mplp_func, (void**)data);
 	while (bam_mplp_auto(iter, &tid, &pos, n_plp, plp) > 0) {
 		if (conf->reg && (pos < beg0 || pos >= end0)) continue; // out of the region requested
+		if (hash) {
+			khint_t k;
+			k = kh_get(64, hash, (uint64_t)tid<<32 | pos);
+			if (k == kh_end(hash)) continue;
+		}
 		if (tid != ref_tid) {
 			free(ref);
 			if (conf->fai) ref = fai_fetch(conf->fai, h->target_name[tid], &ref_len);
 			ref_tid = tid;
 		}
-		printf("%s\t%d\t%c", h->target_name[tid], pos + 1, (ref && pos < ref_len)? ref[pos] : 'N');
-		for (i = 0; i < n; ++i) {
-			int j;
-			printf("\t%d\t", n_plp[i]);
-			if (n_plp[i] == 0) printf("*\t*");
-			else {
-				for (j = 0; j < n_plp[i]; ++j)
-					pileup_seq(plp[i] + j, pos, ref_len, ref);
-				putchar('\t');
-				for (j = 0; j < n_plp[i]; ++j) {
-					const bam_pileup1_t *p = plp[i] + j;
-					int c = bam1_qual(p->b)[p->qpos] + 33;
-					if (c > 126) c = 126;
-					putchar(c);
+		if (conf->flag & MPLP_GLF) {
+			int _ref0, ref16;
+			bcf1_t *b = calloc(1, sizeof(bcf1_t));
+			_ref0 = (ref && pos < ref_len)? ref[pos] : 'N';
+			ref16 = bam_nt16_table[_ref0];
+			for (i = 0; i < n; ++i)
+				bcf_call_glfgen(n_plp[i], plp[i], ref16, bca, bcr + i);
+			bcf_call_combine(n, bcr, ref16, &bc);
+			bcf_call2bcf(tid, pos, &bc, b);
+			bcf_write(bp, bh, b);
+			//fprintf(stderr, "%d,%d,%d\n", b->tid, b->pos, b->l_str);
+			bcf_destroy(b);
+		} else {
+			printf("%s\t%d\t%c", h->target_name[tid], pos + 1, (ref && pos < ref_len)? ref[pos] : 'N');
+			for (i = 0; i < n; ++i) {
+				int j;
+				printf("\t%d\t", n_plp[i]);
+				if (n_plp[i] == 0) printf("*\t*");
+				else {
+					for (j = 0; j < n_plp[i]; ++j)
+						pileup_seq(plp[i] + j, pos, ref_len, ref);
+					putchar('\t');
+					for (j = 0; j < n_plp[i]; ++j) {
+						const bam_pileup1_t *p = plp[i] + j;
+						int c = bam1_qual(p->b)[p->qpos] + 33;
+						if (c > 126) c = 126;
+						putchar(c);
+					}
 				}
 			}
+			putchar('\n');
 		}
-		putchar('\n');
 	}
+	bcf_close(bp);
+	if (hash) { // free the hash table
+		khint_t k;
+		for (k = kh_begin(hash); k < kh_end(hash); ++k)
+			if (kh_exist(hash, k)) free(kh_val(hash, k));
+		kh_destroy(64, hash);
+	}
+	bcf_hdr_destroy(bh); bcf_call_destroy(bca); free(bc.PL); free(bcr);
 	bam_mplp_destroy(iter);
 	bam_header_destroy(h);
 	for (i = 0; i < n; ++i) {
@@ -550,17 +632,38 @@ int bam_mpileup(int argc, char *argv[])
 	int c;
 	mplp_conf_t mplp;
 	memset(&mplp, 0, sizeof(mplp_conf_t));
-	while ((c = getopt(argc, argv, "f:r:")) >= 0) {
+	mplp.max_mq = 60;
+	mplp.theta = 1e-3;
+	mplp.min_baseQ = 13;
+	while ((c = getopt(argc, argv, "gf:r:l:M:q:t:Q:u")) >= 0) {
 		switch (c) {
+		case 't': mplp.theta = atof(optarg); break;
 		case 'f':
 			mplp.fai = fai_load(optarg);
 			if (mplp.fai == 0) return 1;
 			break;
-		case 'r': mplp.reg = strdup(optarg);
+		case 'r': mplp.reg = strdup(optarg); break;
+		case 'l': mplp.fn_pos = strdup(optarg); break;
+		case 'g': mplp.flag |= MPLP_GLF; break;
+		case 'u': mplp.flag |= MPLP_NO_COMP; break;
+		case 'M': mplp.max_mq = atoi(optarg); break;
+		case 'q': mplp.min_mq = atoi(optarg); break;
+		case 'Q': mplp.min_baseQ = atoi(optarg); break;
 		}
 	}
 	if (argc == 1) {
-		fprintf(stderr, "Usage: samtools mpileup [-r reg] [-f in.fa] in1.bam [in2.bam [...]]\n");
+		fprintf(stderr, "\n");
+		fprintf(stderr, "Usage:   samtools mpileup [options] in1.bam [in2.bam [...]]\n\n");
+		fprintf(stderr, "Options: -f FILE     reference sequence file [null]\n");
+		fprintf(stderr, "         -r STR      region in which pileup is generated [null]\n");
+		fprintf(stderr, "         -l FILE     list of positions (format: chr pos) [null]\n");
+		fprintf(stderr, "         -M INT      cap mapping quality at INT [%d]\n", mplp.max_mq);
+		fprintf(stderr, "         -Q INT      min base quality [%d]\n", mplp.min_baseQ);
+		fprintf(stderr, "         -q INT      filter out alignment with MQ smaller than INT [%d]\n", mplp.min_mq);
+		fprintf(stderr, "         -t FLOAT    scaled mutation rate [%lg]\n", mplp.theta);
+		fprintf(stderr, "         -g          generate GLF output\n");
+		fprintf(stderr, "\n");
+		fprintf(stderr, "Notes: Assuming diploid individuals.\n\n");
 		return 1;
 	}
 	mpileup(&mplp, argc - optind, argv + optind);