works

[samtools.git] / bam2bcf.c

diff --git a/bam2bcf.c b/bam2bcf.c
index dec3305340f689a100d29dc4f9b5998e7692c6d4..a51a406f249c05834b0d8e20f49403d5ed29a3b6 100644 (file)
--- a/bam2bcf.c
+++ b/bam2bcf.c
@@ -26,6 +26,7 @@ bcf_callaux_t *bcf_call_init(double theta, int min_baseQ)
        bca->e = errmod_init(1. - theta);
        bca->min_frac = 0.002;
        bca->min_support = 1;
+    bca->per_sample_flt = 0;
        return bca;
 }
 
@@ -54,8 +55,7 @@ int bcf_call_glfgen(int _n, const bam_pileup1_t *pl, int ref_base, bcf_callaux_t
                bca->bases = (uint16_t*)realloc(bca->bases, 2 * bca->max_bases);
        }
        // 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
@@ -78,6 +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;
@@ -86,9 +87,9 @@ int bcf_call_glfgen(int _n, const bam_pileup1_t *pl, int ref_base, bcf_callaux_t
                if (min_dist > p->qpos) min_dist = p->qpos;
                if (min_dist > CAP_DIST) min_dist = CAP_DIST;
                r->anno[1<<2|is_diff<<1|0] += baseQ;
-               r->anno[1<<2|is_diff<<1|1] += baseQ * baseQ;
+               r->anno[1<<2|is_diff<<1|1] += baseQ * baseQ;    // FIXME: signed int is not enough for thousands of samples
                r->anno[2<<2|is_diff<<1|0] += mapQ;
-               r->anno[2<<2|is_diff<<1|1] += mapQ * mapQ;
+               r->anno[2<<2|is_diff<<1|1] += mapQ * mapQ;              // FIXME: signed int is not enough for thousands of samples
                r->anno[3<<2|is_diff<<1|0] += min_dist;
                r->anno[3<<2|is_diff<<1|1] += min_dist * min_dist;
        }
@@ -194,6 +195,8 @@ int bcf_call_combine(int n, const bcf_callret1_t *calls, int ref_base /*4-bit*/,
        for (i = 0; i < n; ++i)
                for (j = 0; j < 4; ++j)
                        qsum[j] += calls[i].qsum[j];
+    int qsum_tot=0;
+    for (j=0; j<4; j++) { qsum_tot += qsum[j]; call->qsum[j] = 0; }
        for (j = 0; j < 4; ++j) qsum[j] = qsum[j] << 2 | j;
        // find the top 2 alleles
        for (i = 1; i < 4; ++i) // insertion sort
@@ -205,9 +208,15 @@ int bcf_call_combine(int n, const bcf_callret1_t *calls, int ref_base /*4-bit*/,
        call->a[0] = ref4;
        for (i = 3, j = 1; i >= 0; --i) {
                if ((qsum[i]&3) != ref4) {
-                       if (qsum[i]>>2 != 0) call->a[j++] = qsum[i]&3;
+                       if (qsum[i]>>2 != 0) 
+            {
+                if ( j<4 ) call->qsum[j] = (float)(qsum[i]>>2)/qsum_tot; // ref N can make j>=4
+                call->a[j++]  = qsum[i]&3;
+            }
                        else break;
                }
+        else 
+            call->qsum[0] = (float)(qsum[i]>>2)/qsum_tot;
        }
        if (ref_base >= 0) { // for SNPs, find the "unseen" base
                if (((ref4 < 4 && j < 4) || (ref4 == 4 && j < 5)) && i >= 0)
@@ -260,7 +269,7 @@ int bcf_call_combine(int n, const bcf_callret1_t *calls, int ref_base /*4-bit*/,
        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);
@@ -304,34 +313,36 @@ 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);
+       if (bc->ori_ref < 0) ksprintf(&s,"INDEL;IS=%d,%f;", bca->max_support, bca->max_frac);
        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);
        }
-    if ( bc->vdb!=1 )
-    {
+    ksprintf(&s,";QS=%f,%f,%f,%f", bc->qsum[0],bc->qsum[1],bc->qsum[2],bc->qsum[3]);
+    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;
-               int32_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)
                                {