X-Git-Url: https://git.donarmstrong.com/?a=blobdiff_plain;f=bam_plcmd.c;h=bbdca64d4cc38ac4edde1691f527a886880eac2b;hb=7f8bb285b631bd30a116ff5b5495563c7f5438f2;hp=708e63e7e8b600baf166e23979990711d9bf5829;hpb=7795f54fa8aed1e0b1edff61890e9fe25fdb7fe9;p=samtools.git

diff --git a/bam_plcmd.c b/bam_plcmd.c
index 708e63e..bbdca64 100644
--- a/bam_plcmd.c
+++ b/bam_plcmd.c
@@ -449,8 +449,16 @@ int bam_pileup(int argc, char *argv[])
  * mpileup *
  ***********/
 
+#define MPLP_VCF   0x1
+#define MPLP_VAR   0x2
+#define MPLP_AFS   0x4
+#define MPLP_AFALL 0x8
+
+#define MPLP_AFS_BLOCK 0x10000
+
 typedef struct {
-	int vcf, max_mq, min_mq;
+	int max_mq, min_mq, prior_type, flag;
+	double theta;
 	char *reg, *fn_pos;
 	faidx_t *fai;
 	kh_64_t *hash;
@@ -480,6 +488,7 @@ static int mpileup(mplp_conf_t *conf, int n, char **fn)
 	const bam_pileup1_t **plp;
 	bam_mplp_t iter;
 	bam_header_t *h = 0;
+	uint64_t N = 0;
 	char *ref;
 	khash_t(64) *hash = 0;
 	// allocate
@@ -517,9 +526,14 @@ static int mpileup(mplp_conf_t *conf, int n, char **fn)
 	}
 	if (conf->fn_pos) hash = load_pos(conf->fn_pos, h);
 	// write the VCF header
-	if (conf->vcf) {
+	if (conf->flag & MPLP_VCF) {
 		kstring_t s;
 		s.l = s.m = 0; s.s = 0;
+		puts("##fileformat=VCFv4.0");
+		puts("##INFO=<ID=DP,Number=1,Type=Integer,Description=\"Total read depth\">");
+		puts("##INFO=<ID=AF,Number=1,Type=Float,Description=\"Non-reference allele frequency \\argmax_f P(D|f)\">");
+		puts("##INFO=<ID=AFE,Number=1,Type=Float,Description=\"Expected non-reference allele frequency\">");
+		puts("##FILTER=<ID=Q13,Description=\"All min{baseQ,mapQ} below 13\">");
 		kputs("#CHROM\tPOS\tID\tREF\tALT\tQUAL\tFILTER\tINFO\tFORMAT", &s);
 		for (i = 0; i < n; ++i) {
 			const char *p;
@@ -532,7 +546,10 @@ static int mpileup(mplp_conf_t *conf, int n, char **fn)
 		free(s.s);
 	}
 	// mpileup
-	if (conf->vcf) ma = mc_init(n);
+	if (conf->flag & MPLP_VCF) {
+		ma = mc_init(n);
+		mc_init_prior(ma, conf->prior_type, conf->theta);
+	}
 	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) {
@@ -547,26 +564,31 @@ static int mpileup(mplp_conf_t *conf, int n, char **fn)
 			if (conf->fai) ref = fai_fetch(conf->fai, h->target_name[tid], &ref_len);
 			ref_tid = tid;
 		}
-		if (conf->vcf) {
-			double f0, f; // the reference allele frequency
-			int j, _ref, _alt, _ref0, depth, rms_q, _ref0b;
+		if (conf->flag & MPLP_VCF) {
+			mc_rst_t r;
+			int j, _ref0, depth, rms_q, _ref0b, is_var = 0, qref = 0, level = 2, tot;
 			uint64_t sqr_sum;
+			if (conf->flag & MPLP_AFS) level = 3;
 			_ref0 = _ref0b = (ref && pos < ref_len)? ref[pos] : 'N';
 			_ref0 = bam_nt16_nt4_table[bam_nt16_table[_ref0]];
-			f = f0 = mc_freq0(_ref0, n_plp, plp, ma, &_ref, &_alt);
-			if (f >= 0.0) {
-				double flast = f;
-				for (j = 0; j < 10; ++j) {
-					f = mc_freq_iter(flast, ma);
-					if (fabs(f - flast) < 1e-3) break;
-					flast = f;
-				}
+			tot = mc_cal(_ref0, n_plp, plp, ma, &r, level);
+			if (tot) { // has good bases
+				double q;
+				is_var = (r.p_ref < .5);
+				q = is_var? r.p_ref : 1. - r.p_ref;
+				if (q < 1e-308) q = 1e-308;
+				qref = (int)(-3.434 * log(q) + .499);
+				if (qref > 99) qref = 99;
 			}
+			if ((conf->flag & MPLP_VAR) && !is_var) continue;
+			++N; // number of processed lines
 			printf("%s\t%d\t.\t%c\t", h->target_name[tid], pos + 1, _ref0b);
-			if (_ref0 == _ref) putchar("ACGTN"[_alt]);
-			else printf("%c,%c", "ACGTN"[_ref], "ACGTN"[_alt]);
-			printf("\t0\t"); // FIXME: currently these not available
-			if (f0 < 0.) printf("Q13\t");
+			if (is_var) {
+				putchar("ACGTN"[r.alt]);
+				if (r.alt2 >= 0 && r.alt2 < 4) printf(",%c", "ACGT"[r.alt2]);
+			} else putchar('.');
+			printf("\t%d\t", qref);
+			if (!tot) printf("Q13\t");
 			else printf(".\t");
 			for (i = depth = 0, sqr_sum = 0; i < n; ++i) {
 				depth += n_plp[i];
@@ -577,12 +599,24 @@ static int mpileup(mplp_conf_t *conf, int n, char **fn)
 				}
 			}
 			rms_q = (int)(sqrt((double)sqr_sum / depth) + .499);
-			printf("DP=%d;MQ=%d;AF=%.3lg", depth, rms_q, f0<0.?0.:1.-f);
-			printf("\tDP");
-			// output genotype information; FIXME: to be implmented...
-			for (i = 0; i < n; ++i)
-				printf("\t%d", n_plp[i]);
+			printf("DP=%d;MQ=%d", depth, rms_q);
+			if (tot) {
+				printf(";AF=%.3lf", 1. - r.f_em);
+				if (level >= 2) printf(";AFE=%.3lf", 1-r.f_exp);
+				if (conf->flag & MPLP_AFALL) {
+					printf(";AF0=%.3lf;AFN=%.3lf", 1-r.f_naive, 1-r.f_nielsen);
+					if (conf->flag & MPLP_AFS) printf(";AFB=%.3lf", 1-r.f_map);
+				}
+			}
+			printf("\tGT:GQ:DP");
+			if (tot) {
+				for (i = 0; i < n; ++i) {
+					int x = mc_call_gt(ma, r.f_exp, i);
+					printf("\t%c/%c:%d:%d", "10"[((x&3)==2)], "10"[((x&3)>0)], x>>2, n_plp[i]);
+				}
+			} else for (i = 0; i < n; ++i) printf("\t./.:0:0");
 			putchar('\n');
+			if (N % MPLP_AFS_BLOCK == 0) mc_dump_afs(ma);
 		} 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) {
@@ -604,6 +638,7 @@ static int mpileup(mplp_conf_t *conf, int n, char **fn)
 			putchar('\n');
 		}
 	}
+	if ((conf->flag&MPLP_VCF) && (conf->flag&MPLP_AFS)) mc_dump_afs(ma);
 	if (hash) { // free the hash table
 		khint_t k;
 		for (k = kh_begin(hash); k < kh_end(hash); ++k)
@@ -628,21 +663,50 @@ int bam_mpileup(int argc, char *argv[])
 	mplp_conf_t mplp;
 	memset(&mplp, 0, sizeof(mplp_conf_t));
 	mplp.max_mq = 60;
-	while ((c = getopt(argc, argv, "f:r:l:VM:q:")) >= 0) {
+	mplp.prior_type = MC_PTYPE_FULL;
+	mplp.theta = 1e-3;
+	while ((c = getopt(argc, argv, "vVcFSP:f:r:l:VM:q:t:")) >= 0) {
 		switch (c) {
+		case 't': mplp.theta = atof(optarg); break;
+		case 'P':
+			if (strcmp(optarg, "full") == 0) mplp.prior_type = MC_PTYPE_FULL;
+			else if (strcmp(optarg, "cond2") == 0) mplp.prior_type = MC_PTYPE_COND2;
+			else if (strcmp(optarg, "flat") == 0) mplp.prior_type = MC_PTYPE_FLAT;
+			else {
+				fprintf(stderr, "[%s] unrecognized prior type.\n", __func__);
+				return 1;
+			}
+			break;
 		case 'f':
 			mplp.fai = fai_load(optarg);
 			if (mplp.fai == 0) return 1;
 			break;
 		case 'r': mplp.reg = strdup(optarg); break;
 		case 'l': mplp.fn_pos = strdup(optarg); break;
-		case 'V': mplp.vcf = 1; break;
+		case 'V':
+		case 'c': mplp.flag |= MPLP_VCF; break;
+		case 'S': mplp.flag |= MPLP_AFS; break;
+		case 'F': mplp.flag |= MPLP_AFALL; break;
+		case 'v': mplp.flag |= MPLP_VAR; break;
 		case 'M': mplp.max_mq = atoi(optarg); break;
 		case 'q': mplp.min_mq = atoi(optarg); break;
 		}
 	}
 	if (argc == 1) {
-		fprintf(stderr, "Usage: samtools mpileup [-r reg] [-f in.fa] [-l pos] 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      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, "         -P STR      prior: full, flat, cond2 [full]\n");
+		fprintf(stderr, "         -c          generate VCF output (consensus calling)\n");
+		fprintf(stderr, "         -v          show variant sites only\n");
+		fprintf(stderr, "         -S          calculate AFS (slow, to stderr)\n");
+		fprintf(stderr, "\n");
+		fprintf(stderr, "Notes: Assuming error independency and diploid individuals.\n\n");
 		return 1;
 	}
 	mpileup(&mplp, argc - optind, argv + optind);