X-Git-Url: https://git.donarmstrong.com/?a=blobdiff_plain;f=bam_plcmd.c;h=7f13eee4ed51a987a6386796eb4eeae2bad65891;hb=4c8c9dfc1e3b3b066a62a703fd3ba04db6ad5a45;hp=a3e6aeb614927733676b2b25e4592786cb14ef2e;hpb=2d2be1edb0548e95b771bc9c79cff48eb5d2e1f5;p=samtools.git

diff --git a/bam_plcmd.c b/bam_plcmd.c
index a3e6aeb..7f13eee 100644
--- a/bam_plcmd.c
+++ b/bam_plcmd.c
@@ -532,10 +532,12 @@ int bam_pileup(int argc, char *argv[])
 #define MPLP_REALN   0x80
 #define MPLP_FMT_DP 0x100
 #define MPLP_FMT_SP 0x200
+#define MPLP_NO_INDEL 0x400
 
 typedef struct {
 	int max_mq, min_mq, flag, min_baseQ, capQ_thres, max_depth;
-	char *reg, *fn_pos;
+	int openQ, extQ, tandemQ;
+	char *reg, *fn_pos, *pl_list;
 	faidx_t *fai;
 	kh_64_t *hash;
 } mplp_conf_t;
@@ -556,7 +558,7 @@ typedef struct {
 static int mplp_func(void *data, bam1_t *b)
 {
 	extern int bam_realn(bam1_t *b, const char *ref);
-	extern int bam_prob_realn(bam1_t *b, const char *ref);
+	extern int bam_prob_realn_core(bam1_t *b, const char *ref, int);
 	extern int bam_cap_mapQ(bam1_t *b, char *ref, int thres);
 	mplp_aux_t *ma = (mplp_aux_t*)data;
 	int ret, skip = 0;
@@ -565,7 +567,7 @@ static int mplp_func(void *data, bam1_t *b)
 		ret = ma->iter? bam_iter_read(ma->fp, ma->iter, b) : bam_read1(ma->fp, b);
 		if (ret < 0) break;
 		skip = 0;
-		if (has_ref && (ma->flag&MPLP_REALN)) bam_prob_realn(b, ma->ref);
+		if (has_ref && (ma->flag&MPLP_REALN)) bam_prob_realn_core(b, ma->ref, 1);
 		if (has_ref && ma->capQ_thres > 10) {
 			int q = bam_cap_mapQ(b, ma->ref, ma->capQ_thres);
 			if (q < 0) skip = 1;
@@ -602,6 +604,8 @@ static void group_smpl(mplp_pileup_t *m, bam_sample_t *sm, kstring_t *buf,
 
 static int mpileup(mplp_conf_t *conf, int n, char **fn)
 {
+	extern void *bcf_call_add_rg(void *rghash, const char *hdtext, const char *list);
+	extern void bcf_call_del_rghash(void *rghash);
 	mplp_aux_t **data;
 	int i, tid, pos, *n_plp, beg0 = 0, end0 = 1u<<29, ref_len, ref_tid, max_depth;
 	const bam_pileup1_t **plp;
@@ -609,6 +613,7 @@ static int mpileup(mplp_conf_t *conf, int n, char **fn)
 	bam_header_t *h = 0;
 	char *ref;
 	khash_t(64) *hash = 0;
+	void *rghash = 0;
 
 	bcf_callaux_t *bca = 0;
 	bcf_callret1_t *bcr = 0;
@@ -638,6 +643,7 @@ static int mpileup(mplp_conf_t *conf, int n, char **fn)
 		data[i]->fp = strcmp(fn[i], "-") == 0? bam_dopen(fileno(stdin), "r") : bam_open(fn[i], "r");
 		h_tmp = bam_header_read(data[i]->fp);
 		bam_smpl_add(sm, fn[i], h_tmp->text);
+		rghash = bcf_call_add_rg(rghash, h_tmp->text, conf->pl_list);
 		if (conf->reg) {
 			int beg, end;
 			bam_index_t *idx;
@@ -693,6 +699,8 @@ static int mpileup(mplp_conf_t *conf, int n, char **fn)
 		bcf_hdr_write(bp, bh);
 		bca = bcf_call_init(-1., conf->min_baseQ);
 		bcr = calloc(sm->n, sizeof(bcf_callret1_t));
+		bca->rghash = rghash;
+		bca->openQ = conf->openQ, bca->extQ = conf->extQ, bca->tandemQ = conf->tandemQ;
 	}
 	ref_tid = -1; ref = 0;
 	iter = bam_mplp_init(n, mplp_func, (void**)data);
@@ -726,9 +734,37 @@ static int mpileup(mplp_conf_t *conf, int n, char **fn)
 			for (i = 0; i < gplp.n; ++i)
 				bcf_call_glfgen(gplp.n_plp[i], gplp.plp[i], ref16, bca, bcr + i);
 			bcf_call_combine(gplp.n, bcr, ref16, &bc);
-			bcf_call2bcf(tid, pos, &bc, b, (conf->flag&(MPLP_FMT_DP|MPLP_FMT_SP))? bcr : 0, (conf->flag&MPLP_FMT_SP));
+			bcf_call2bcf(tid, pos, &bc, b, (conf->flag&(MPLP_FMT_DP|MPLP_FMT_SP))? bcr : 0,
+						 (conf->flag&MPLP_FMT_SP), 0, 0);
 			bcf_write(bp, bh, b);
 			bcf_destroy(b);
+			if (!(conf->flag&MPLP_NO_INDEL)) {
+				// call MNPs
+				if (bcf_call_mnp_prep(gplp.n, gplp.n_plp, gplp.plp, pos, bca, ref) >= 0) {
+					for (i = 0; i < gplp.n; ++i)
+						bcf_call_glfgen(gplp.n_plp[i], gplp.plp[i], B2B_REF_MNP, bca, bcr + i);
+					if (bcf_call_combine(gplp.n, bcr, B2B_REF_MNP, &bc) >= 0) {
+						b = calloc(1, sizeof(bcf1_t));
+						bcf_call2bcf(tid, pos, &bc, b, (conf->flag&(MPLP_FMT_DP|MPLP_FMT_SP))? bcr : 0,
+									 (conf->flag&MPLP_FMT_SP), bca, ref);
+						bcf_write(bp, bh, b);
+						bcf_destroy(b);
+						bca->last_mnp_pos = pos;
+					}
+				}
+				// call indels
+				if (bcf_call_gap_prep(gplp.n, gplp.n_plp, gplp.plp, pos, bca, ref, rghash) >= 0) {
+					for (i = 0; i < gplp.n; ++i)
+						bcf_call_glfgen(gplp.n_plp[i], gplp.plp[i], B2B_REF_INDEL, bca, bcr + i);
+					if (bcf_call_combine(gplp.n, bcr, B2B_REF_INDEL, &bc) >= 0) {
+						b = calloc(1, sizeof(bcf1_t));
+						bcf_call2bcf(tid, pos, &bc, b, (conf->flag&(MPLP_FMT_DP|MPLP_FMT_SP))? bcr : 0,
+									 (conf->flag&MPLP_FMT_SP), bca, ref);
+						bcf_write(bp, bh, b);
+						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) {
@@ -755,6 +791,7 @@ static int mpileup(mplp_conf_t *conf, int n, char **fn)
 	bam_smpl_destroy(sm); free(buf.s);
 	for (i = 0; i < gplp.n; ++i) free(gplp.plp[i]);
 	free(gplp.plp); free(gplp.n_plp); free(gplp.m_plp);
+	bcf_call_del_rghash(rghash);
 	if (hash) { // free the hash table
 		khint_t k;
 		for (k = kh_begin(hash); k < kh_end(hash); ++k)
@@ -834,8 +871,9 @@ int bam_mpileup(int argc, char *argv[])
 	mplp.min_baseQ = 13;
 	mplp.capQ_thres = 0;
 	mplp.max_depth = 250;
+	mplp.openQ = 40; mplp.extQ = 20; mplp.tandemQ = 100;
 	mplp.flag = MPLP_NO_ORPHAN | MPLP_REALN;
-	while ((c = getopt(argc, argv, "gf:r:l:M:q:Q:uaORC:BDSd:b:")) >= 0) {
+	while ((c = getopt(argc, argv, "gf:r:l:M:q:Q:uaORC:BDSd:b:P:o:e:h:I")) >= 0) {
 		switch (c) {
 		case 'f':
 			mplp.fai = fai_load(optarg);
@@ -844,6 +882,7 @@ int bam_mpileup(int argc, char *argv[])
 		case 'd': mplp.max_depth = atoi(optarg); break;
 		case 'r': mplp.reg = strdup(optarg); break;
 		case 'l': mplp.fn_pos = strdup(optarg); break;
+		case 'P': mplp.pl_list = strdup(optarg); break;
 		case 'g': mplp.flag |= MPLP_GLF; break;
 		case 'u': mplp.flag |= MPLP_NO_COMP | MPLP_GLF; break;
 		case 'a': mplp.flag |= MPLP_NO_ORPHAN | MPLP_REALN; break;
@@ -852,11 +891,15 @@ int bam_mpileup(int argc, char *argv[])
 		case 'R': mplp.flag |= MPLP_REALN; break;
 		case 'D': mplp.flag |= MPLP_FMT_DP; break;
 		case 'S': mplp.flag |= MPLP_FMT_SP; break;
+		case 'I': mplp.flag |= MPLP_NO_INDEL; break;
 		case 'C': mplp.capQ_thres = atoi(optarg); 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;
         case 'b': file_list = optarg; break;
+		case 'o': mplp.openQ = atoi(optarg); break;
+		case 'e': mplp.extQ = atoi(optarg); break;
+		case 'h': mplp.tandemQ = atoi(optarg); break;
 		}
 	}
 	if (argc == 1) {
@@ -870,11 +913,16 @@ int bam_mpileup(int argc, char *argv[])
 		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, "         -d INT      max per-sample depth [%d]\n", mplp.max_depth);
+		fprintf(stderr, "         -P STR      comma separated list of platforms for indels [all]\n");
+		fprintf(stderr, "         -o INT      Phred-scaled gap open sequencing error probability [%d]\n", mplp.openQ);
+		fprintf(stderr, "         -e INT      Phred-scaled gap extension seq error probability [%d]\n", mplp.extQ);
+		fprintf(stderr, "         -h INT      coefficient for homopolyer errors [%d]\n", mplp.tandemQ);
 		fprintf(stderr, "         -g          generate BCF output\n");
 		fprintf(stderr, "         -u          do not compress BCF output\n");
 		fprintf(stderr, "         -B          disable BAQ computation\n");
 		fprintf(stderr, "         -D          output per-sample DP\n");
 		fprintf(stderr, "         -S          output per-sample SP (strand bias P-value, slow)\n");
+		fprintf(stderr, "         -I          do not perform indel calling\n");
 		fprintf(stderr, "\n");
 		fprintf(stderr, "Notes: Assuming diploid individuals.\n\n");
 		return 1;
@@ -888,7 +936,7 @@ int bam_mpileup(int argc, char *argv[])
     }
     else
 	    mpileup(&mplp, argc - optind, argv + optind);
-	free(mplp.reg);
+	free(mplp.reg); free(mplp.pl_list);
 	if (mplp.fai) fai_destroy(mplp.fai);
 	return 0;
 }