X-Git-Url: https://git.donarmstrong.com/?a=blobdiff_plain;f=bam2bcf.c;h=6ac5dce182a2353ada163dd65bdb5ca530d59127;hb=9f118264ea012adc21a46d7c03eaad4f9ce7d4d4;hp=bd5503d7e55cef2479c1a8d12d3dc0855eab06f2;hpb=27eec65490fb3461f0af054868d9a0949e1843ed;p=samtools.git

diff --git a/bam2bcf.c b/bam2bcf.c
index bd5503d..6ac5dce 100644
--- a/bam2bcf.c
+++ b/bam2bcf.c
@@ -39,6 +39,7 @@ void bcf_call_destroy(bcf_callaux_t *bca)
  * negative if we are looking at an indel. */
 int bcf_call_glfgen(int _n, const bam_pileup1_t *pl, int ref_base, bcf_callaux_t *bca, bcf_callret1_t *r)
 {
+    static int *var_pos = NULL, nvar_pos = 0;
 	int i, n, ref4, is_indel, ori_depth = 0;
 	memset(r, 0, sizeof(bcf_callret1_t));
 	if (ref_base >= 0) {
@@ -54,7 +55,7 @@ int bcf_call_glfgen(int _n, const bam_pileup1_t *pl, int ref_base, bcf_callaux_t
 	}
 	// fill the bases array
 	memset(r, 0, sizeof(bcf_callret1_t));
-	for (i = n = 0; i < _n; ++i) {
+	for (i = n = r->n_supp = 0; i < _n; ++i) {
 		const bam_pileup1_t *p = pl + i;
 		int q, b, mapQ, baseQ, is_diff, min_dist, seqQ;
 		// set base
@@ -77,6 +78,7 @@ int bcf_call_glfgen(int _n, const bam_pileup1_t *pl, int ref_base, bcf_callaux_t
 			b = p->aux>>16&0x3f;
 			is_diff = (b != 0);
 		}
+		if (is_diff) ++r->n_supp;
 		bca->bases[n++] = q<<5 | (int)bam1_strand(p->b)<<4 | b;
 		// collect annotations
 		if (b < 4) r->qsum[b] += q;
@@ -94,9 +96,92 @@ int bcf_call_glfgen(int _n, const bam_pileup1_t *pl, int ref_base, bcf_callaux_t
 	r->depth = n; r->ori_depth = ori_depth;
 	// glfgen
 	errmod_cal(bca->e, n, 5, bca->bases, r->p);
+
+    // Calculate the Variant Distance Bias (make it optional?)
+    if ( nvar_pos < _n ) {
+        nvar_pos = _n;
+        var_pos = realloc(var_pos,sizeof(int)*nvar_pos);
+    }
+    int alt_dp=0, read_len=0;
+    for (i=0; i<_n; i++) {
+        const bam_pileup1_t *p = pl + i;
+        if ( bam1_seqi(bam1_seq(p->b),p->qpos) == ref_base ) 
+            continue;
+
+        var_pos[alt_dp] = p->qpos;
+        if ( (bam1_cigar(p->b)[0]&BAM_CIGAR_MASK)==4 )
+            var_pos[alt_dp] -= bam1_cigar(p->b)[0]>>BAM_CIGAR_SHIFT;
+
+        alt_dp++;
+        read_len += p->b->core.l_qseq;
+    }
+    float mvd=0;
+    int j;
+    n=0;
+    for (i=0; i<alt_dp; i++) {
+        for (j=0; j<i; j++) {
+            mvd += abs(var_pos[i] - var_pos[j]);
+            n++;
+        }
+    }
+    r->mvd[0] = n ? mvd/n : 0;
+    r->mvd[1] = alt_dp;
+    r->mvd[2] = alt_dp ? read_len/alt_dp : 0;
+
 	return r->depth;
 }
 
+
+void calc_vdb(int n, const bcf_callret1_t *calls, bcf_call_t *call)
+{
+    // Variant distance bias. Samples merged by means of DP-weighted average.
+
+    float weight=0, tot_prob=0;
+
+    int i;
+    for (i=0; i<n; i++)
+    {
+        int mvd      = calls[i].mvd[0];
+        int dp       = calls[i].mvd[1];
+        int read_len = calls[i].mvd[2];
+
+        if ( dp<2 ) continue;
+
+        float prob = 0;
+        if ( dp==2 )
+        {
+            // Exact formula
+            prob = (mvd==0) ? 1.0/read_len : (read_len-mvd)*2.0/read_len/read_len;
+        }
+        else if ( dp==3 )
+        {
+            // Sin, quite accurate approximation
+            float mu = read_len/2.9;
+            prob = mvd>2*mu ? 0 : sin(mvd*3.14/2/mu) / (4*mu/3.14);
+        }
+        else
+        {
+            // Scaled gaussian curve, crude approximation, but behaves well. Using fixed depth for bigger depths.
+            if ( dp>5 )
+                dp = 5;
+            float sigma2 = (read_len/1.9/(dp+1)) * (read_len/1.9/(dp+1));
+            float norm   = 1.125*sqrt(2*3.14*sigma2);
+            float mu     = read_len/2.9;
+            if ( mvd < mu )
+                prob = exp(-(mvd-mu)*(mvd-mu)/2/sigma2)/norm;
+            else
+                prob = exp(-(mvd-mu)*(mvd-mu)/3.125/sigma2)/norm;
+        }
+
+        //fprintf(stderr,"dp=%d mvd=%d read_len=%d -> prob=%f\n", dp,mvd,read_len,prob);
+        tot_prob += prob*dp;
+        weight += dp;
+    }
+    tot_prob = weight ? tot_prob/weight : 1; 
+    //fprintf(stderr,"prob=%f\n", tot_prob);
+    call->vdb = tot_prob;
+}
+
 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];
@@ -170,10 +255,13 @@ int bcf_call_combine(int n, const bcf_callret1_t *calls, int ref_base /*4-bit*/,
 		call->ori_depth += calls[i].ori_depth;
 		for (j = 0; j < 16; ++j) call->anno[j] += calls[i].anno[j];
 	}
+
+    calc_vdb(n, calls, call);
+
 	return 0;
 }
 
-int bcf_call2bcf(int tid, int pos, bcf_call_t *bc, bcf1_t *b, bcf_callret1_t *bcr, int is_SP,
+int bcf_call2bcf(int tid, int pos, bcf_call_t *bc, bcf1_t *b, bcf_callret1_t *bcr, int fmt_flag,
 				 const bcf_callaux_t *bca, const char *ref)
 {
 	extern double kt_fisher_exact(int n11, int n12, int n21, int n22, double *_left, double *_right, double *two);
@@ -223,24 +311,29 @@ int bcf_call2bcf(int tid, int pos, bcf_call_t *bc, bcf1_t *b, bcf_callret1_t *bc
 		if (i) kputc(',', &s);
 		kputw(bc->anno[i], &s);
 	}
+    if (bc->vdb != 1)
+        ksprintf(&s, ";VDB=%.4f", bc->vdb);
 	kputc('\0', &s);
 	// FMT
 	kputs("PL", &s);
-	if (bcr) {
-		kputs(":DP", &s);
-		if (is_SP) kputs(":SP", &s);
+	if (bcr && fmt_flag) {
+		if (fmt_flag & B2B_FMT_DP) kputs(":DP", &s);
+		if (fmt_flag & B2B_FMT_DV) kputs(":DV", &s);
+		if (fmt_flag & B2B_FMT_SP) kputs(":SP", &s);
 	}
 	kputc('\0', &s);
 	b->m_str = s.m; b->str = s.s; b->l_str = s.l;
 	bcf_sync(b);
 	memcpy(b->gi[0].data, bc->PL, b->gi[0].len * bc->n);
-	if (bcr) {
-		uint16_t *dp = (uint16_t*)b->gi[1].data;
-		uint8_t *sp = is_SP? b->gi[2].data : 0;
+	if (bcr && fmt_flag) {
+		uint16_t *dp = (fmt_flag & B2B_FMT_DP)? b->gi[1].data : 0;
+		uint16_t *dv = (fmt_flag & B2B_FMT_DV)? b->gi[1 + ((fmt_flag & B2B_FMT_DP) != 0)].data : 0;
+		int32_t  *sp = (fmt_flag & B2B_FMT_SP)? b->gi[1 + ((fmt_flag & B2B_FMT_DP) != 0) + ((fmt_flag & B2B_FMT_DV) != 0)].data : 0;
 		for (i = 0; i < bc->n; ++i) {
 			bcf_callret1_t *p = bcr + i;
-			dp[i] = p->depth < 0xffff? p->depth : 0xffff;
-			if (is_SP) {
+			if (dp) dp[i] = p->depth  < 0xffff? p->depth  : 0xffff;
+			if (dv) dv[i] = p->n_supp < 0xffff? p->n_supp : 0xffff;
+			if (sp) {
 				if (p->anno[0] + p->anno[1] < 2 || p->anno[2] + p->anno[3] < 2
 					|| p->anno[0] + p->anno[2] < 2 || p->anno[1] + p->anno[3] < 2)
 				{