X-Git-Url: https://git.donarmstrong.com/?a=blobdiff_plain;f=bam2bcf.c;h=de414d7176903dd323ced99f3b8de39396667517;hb=0735c4ca30725f8f41070890b0e8eb6439846c54;hp=384e481cbc5e6d9d4cae4f2e50a66008f29f1c84;hpb=e2d4caa430112346f387709bd016eb1a8c737d8e;p=samtools.git

diff --git a/bam2bcf.c b/bam2bcf.c
index 384e481..de414d7 100644
--- a/bam2bcf.c
+++ b/bam2bcf.c
@@ -20,8 +20,9 @@ bcf_callaux_t *bcf_call_init(double theta, int min_baseQ)
 	if (theta <= 0.) theta = CALL_DEFTHETA;
 	bca = calloc(1, sizeof(bcf_callaux_t));
 	bca->capQ = 60;
-	bca->openQ = 40; bca->extQ = 20; bca->tandemQ = 80;
+	bca->openQ = 40; bca->extQ = 20; bca->tandemQ = 100;
 	bca->min_baseQ = min_baseQ;
+	bca->last_mnp_pos = -100;
 	bca->e = errmod_init(1. - theta);
 	return bca;
 }
@@ -32,11 +33,13 @@ void bcf_call_destroy(bcf_callaux_t *bca)
 	errmod_destroy(bca->e);
 	free(bca->bases); free(bca->inscns); free(bca);
 }
-/* ref_base is the 4-bit representation of the reference base. It is
- * negative if we are looking at an indel. */
+/* Compute genotype likelihood by combining likelihood of each
+ * read. ref_base is the 4-bit representation of the reference base. It
+ * is negative if we are looking at an indel or an MNP. Indel and MNP
+ * are indistinguishable in this function. */
 int bcf_call_glfgen(int _n, const bam_pileup1_t *pl, int ref_base, bcf_callaux_t *bca, bcf_callret1_t *r)
 {
-	int i, n, ref4, is_indel;
+	int i, n, ref4, is_indel, ori_depth = 0;
 	memset(r, 0, sizeof(bcf_callret1_t));
 	if (ref_base >= 0) {
 		ref4 = bam_nt16_nt4_table[ref_base];
@@ -53,11 +56,14 @@ int bcf_call_glfgen(int _n, const bam_pileup1_t *pl, int ref_base, bcf_callaux_t
 	memset(r, 0, sizeof(bcf_callret1_t));
 	for (i = n = 0; i < _n; ++i) {
 		const bam_pileup1_t *p = pl + i;
-		int q, b, mapQ, baseQ, is_diff, min_dist;
+		int q, b, mapQ, baseQ, is_diff, min_dist, seqQ;
 		// set base
 		if (p->is_del || p->is_refskip || (p->b->core.flag&BAM_FUNMAP)) continue;
+		++ori_depth;
 		baseQ = q = is_indel? p->aux&0xff : (int)bam1_qual(p->b)[p->qpos]; // base/indel quality
+		seqQ = is_indel? (p->aux>>8&0xff) : 99;
 		if (q < bca->min_baseQ) continue;
+		if (q > seqQ) q = seqQ;
 		mapQ = p->b->core.qual < bca->capQ? p->b->core.qual : bca->capQ;
 		if (q > mapQ) q = mapQ;
 		if (q > 63) q = 63;
@@ -67,7 +73,7 @@ int bcf_call_glfgen(int _n, const bam_pileup1_t *pl, int ref_base, bcf_callaux_t
 			b = bam_nt16_nt4_table[b? b : ref_base]; // b is the 2-bit base
 			is_diff = (ref4 < 4 && b == ref4)? 0 : 1;
 		} else {
-			b = p->aux>>8&0x3f;
+			b = p->aux>>16&0x3f;
 			is_diff = (b != 0);
 		}
 		bca->bases[n++] = q<<5 | (int)bam1_strand(p->b)<<4 | b;
@@ -84,12 +90,14 @@ int bcf_call_glfgen(int _n, const bam_pileup1_t *pl, int ref_base, bcf_callaux_t
 		r->anno[3<<2|is_diff<<1|0] += min_dist;
 		r->anno[3<<2|is_diff<<1|1] += min_dist * min_dist;
 	}
-	r->depth = n;
+	r->depth = n; r->ori_depth = ori_depth;
 	// glfgen
 	errmod_cal(bca->e, n, 5, bca->bases, r->p);
 	return r->depth;
 }
-
+/* Combine individual calls. ref_base is a 4-bit A/C/G/T for a SNP, or
+ * B2B_REF_MNP, or B2B_REF_INDEL. Indel and MNP are indistinguishable in
+ * this function. */
 int bcf_call_combine(int n, const bcf_callret1_t *calls, int ref_base /*4-bit*/, bcf_call_t *call)
 {
 	int ref4, i, j, qsum[4];
@@ -97,7 +105,7 @@ int bcf_call_combine(int n, const bcf_callret1_t *calls, int ref_base /*4-bit*/,
 	if (ref_base >= 0) {
 		call->ori_ref = ref4 = bam_nt16_nt4_table[ref_base];
 		if (ref4 > 4) ref4 = 4;
-	} else call->ori_ref = -1, ref4 = 0;
+	} else call->ori_ref = ref_base, ref4 = 0; // ref_base < 0
 	// calculate qsum
 	memset(qsum, 0, 4 * sizeof(int));
 	for (i = 0; i < n; ++i)
@@ -122,7 +130,10 @@ int bcf_call_combine(int n, const bcf_callret1_t *calls, int ref_base /*4-bit*/,
 		if (((ref4 < 4 && j < 4) || (ref4 == 4 && j < 5)) && i >= 0)
 			call->unseen = j, call->a[j++] = qsum[i]&3;
 		call->n_alleles = j;
-	} else call->n_alleles = j;
+	} else { // for indels and MNPs
+		call->n_alleles = j;
+		if (call->n_alleles == 1) return -1; // no reliable supporting read. stop doing anything
+	}
 	// set the PL array
 	if (call->n < n) {
 		call->n = n;
@@ -155,13 +166,15 @@ int bcf_call_combine(int n, const bcf_callret1_t *calls, int ref_base /*4-bit*/,
 	}
 	// combine annotations
 	memset(call->anno, 0, 16 * sizeof(int));
-	for (i = call->depth = 0, tmp = 0; i < n; ++i) {
+	for (i = call->depth = call->ori_depth = 0, tmp = 0; i < n; ++i) {
 		call->depth += calls[i].depth;
+		call->ori_depth += calls[i].ori_depth;
 		for (j = 0; j < 16; ++j) call->anno[j] += calls[i].anno[j];
 	}
 	return 0;
 }
-
+/* Fill a bcf1_t record. The type of the record (SNP, MNP or INDEL) is
+ * determined by bca->ori_ref. */
 int bcf_call2bcf(int tid, int pos, bcf_call_t *bc, bcf1_t *b, bcf_callret1_t *bcr, int is_SP,
 				 const bcf_callaux_t *bca, const char *ref)
 {
@@ -172,7 +185,7 @@ int bcf_call2bcf(int tid, int pos, bcf_call_t *bc, bcf1_t *b, bcf_callret1_t *bc
 	b->tid = tid; b->pos = pos; b->qual = 0;
 	s.s = b->str; s.m = b->m_str; s.l = 0;
 	kputc('\0', &s);
-	if (bc->ori_ref < 0) { // an indel
+	if (bc->ori_ref == B2B_REF_INDEL) { // an indel
 		// write REF
 		kputc(ref[pos], &s);
 		for (j = 0; j < bca->indelreg; ++j) kputc(ref[pos+1+j], &s);
@@ -195,6 +208,16 @@ int bcf_call2bcf(int tid, int pos, bcf_call_t *bc, bcf1_t *b, bcf_callret1_t *bc
 			}
 		}
 		kputc('\0', &s);
+	} else if (bc->ori_ref == B2B_REF_MNP) {
+		for (j = 0; j < bca->indelreg; ++j) kputc(ref[pos+j], &s);
+		kputc('\0', &s);
+		for (i = 1; i < 4; ++i) {
+			if (bc->a[i] < 0) break;
+			if (i > 1) kputc(',', &s);
+			for (j = 0; j < bca->indelreg; ++j)
+				kputc("ACGT"[bca->indel_types[i]>>2*i&3], &s);
+		}
+		kputc('\0', &s);
 	} else { // a SNP
 		kputc("ACGTN"[bc->ori_ref], &s); kputc('\0', &s);
 		for (i = 1; i < 5; ++i) {
@@ -206,8 +229,9 @@ int bcf_call2bcf(int tid, int pos, bcf_call_t *bc, bcf1_t *b, bcf_callret1_t *bc
 	}
 	kputc('\0', &s);
 	// INFO
-	if (bc->ori_ref < 0) kputs("INDEL;", &s);
-	kputs("I16=", &s);
+	if (bc->ori_ref == B2B_REF_INDEL) kputs("INDEL;", &s);
+	else if (bc->ori_ref == B2B_REF_MNP) kputs("MNP;", &s);
+	kputs("DP=", &s); kputw(bc->ori_depth, &s); kputs(";I16=", &s);
 	for (i = 0; i < 16; ++i) {
 		if (i) kputc(',', &s);
 		kputw(bc->anno[i], &s);