fixed a few bugs in the indel caller. Probably there are more.

[samtools.git] / bam2bcf.c

diff --git a/bam2bcf.c b/bam2bcf.c
index 95532254a084caf3fbd6e05840819225848e952b..89d6f9001bfd6dfaed29fa49430e3a23289f58da 100644 (file)
--- a/bam2bcf.c
+++ b/bam2bcf.c
@@ -14,18 +14,13 @@ extern      void ks_introsort_uint32_t(size_t n, uint32_t a[]);
 
 #define CAP_DIST 25
 
-struct __bcf_callaux_t {
-       int max_bases, capQ, min_baseQ;
-       uint16_t *bases;
-       errmod_t *e;
-};
-
 bcf_callaux_t *bcf_call_init(double theta, int min_baseQ)
 {
        bcf_callaux_t *bca;
        if (theta <= 0.) theta = CALL_DEFTHETA;
        bca = calloc(1, sizeof(bcf_callaux_t));
        bca->capQ = 60;
+       bca->openQ = 40; bca->extQ = 20; bca->tandemQ = 100;
        bca->min_baseQ = min_baseQ;
        bca->e = errmod_init(1. - theta);
        return bca;
@@ -154,10 +149,12 @@ 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)
+int bcf_call2bcf(int tid, int pos, bcf_call_t *bc, bcf1_t *b, bcf_callret1_t *bcr, int is_SP)
 {
+       extern double kt_fisher_exact(int n11, int n12, int n21, int n22, double *_left, double *_right, double *two);
        kstring_t s;
        int i;
+       b->n_smpl = bc->n;
        b->tid = tid; b->pos = pos; b->qual = 0;
        s.s = b->str; s.m = b->m_str; s.l = 0;
        kputc('\0', &s);
@@ -177,9 +174,36 @@ int bcf_call2bcf(int tid, int pos, bcf_call_t *bc, bcf1_t *b)
        }
        kputc('\0', &s);
        // FMT
-       kputs("PL", &s); kputc('\0', &s);
+       kputs("PL", &s);
+       if (bcr) {
+               kputs(":DP", &s);
+               if (is_SP) kputs(":SP", &s);
+       }
+       kputc('\0', &s);
        b->m_str = s.m; b->str = s.s; b->l_str = s.l;
-       bcf_sync(bc->n, b);
+       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;
+               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 (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)
+                               {
+                                       sp[i] = 0;
+                               } else {
+                                       double left, right, two;
+                                       int x;
+                                       kt_fisher_exact(p->anno[0], p->anno[1], p->anno[2], p->anno[3], &left, &right, &two);
+                                       x = (int)(-4.343 * log(two) + .499);
+                                       if (x > 255) x = 255;
+                                       sp[i] = x;
+                               }
+                       }
+               }
+       }
        return 0;
 }