Release samtools-0.1.16 (r963:234)

[samtools.git] / bam_maqcns.c

diff --git a/bam_maqcns.c b/bam_maqcns.c
index 9caa5d7b4dfac43d166d163339547a82def69ca3..99310362afa0dfa216b7b678b04f2d69d2103150 100644 (file)
--- a/bam_maqcns.c
+++ b/bam_maqcns.c
@@ -22,8 +22,6 @@ typedef struct {
        uint32_t c[4];
 } glf_call_aux_t;
 
-char bam_nt16_nt4_table[] = { 4, 0, 1, 4, 2, 4, 4, 4, 3, 4, 4, 4, 4, 4, 4, 4 };
-
 /*
   P(<b1,b2>) = \theta \sum_{i=1}^{N-1} 1/i
   P(D|<b1,b2>) = \sum_{k=1}^{N-1} p_k 1/2 [(k/N)^n_2(1-k/N)^n_1 + (k/N)^n1(1-k/N)^n_2]
@@ -110,10 +108,11 @@ bam_maqcns_t *bam_maqcns_init()
        bm = (bam_maqcns_t*)calloc(1, sizeof(bam_maqcns_t));
        bm->aux = (bmc_aux_t*)calloc(1, sizeof(bmc_aux_t));
        bm->het_rate = 0.001;
-       bm->theta = 0.85;
+       bm->theta = 0.83f;
        bm->n_hap = 2;
        bm->eta = 0.03;
        bm->cap_mapQ = 60;
+       bm->min_baseQ = 13;
        return bm;
 }
 
@@ -153,8 +152,9 @@ glf1_t *bam_maqcns_glfgen(int _n, const bam_pileup1_t *pl, uint8_t ref_base, bam
                const bam_pileup1_t *p = pl + i;
                uint32_t q, x = 0, qq;
                uint16_t y = 0;
-               if (p->is_del || (p->b->core.flag&BAM_FUNMAP)) continue;
+               if (p->is_del || p->is_refskip || (p->b->core.flag&BAM_FUNMAP)) continue;
                q = (uint32_t)bam1_qual(p->b)[p->qpos];
+               if (q < bm->min_baseQ) continue;
                x |= (uint32_t)bam1_strand(p->b) << 18 | q << 8 | p->b->core.qual;
                y |= bam1_strand(p->b)<<4;
                if (p->b->core.qual < q) q = p->b->core.qual;
@@ -164,7 +164,7 @@ glf1_t *bam_maqcns_glfgen(int _n, const bam_pileup1_t *pl, uint8_t ref_base, bam
                y |= q << 5;
                qq = bam1_seqi(bam1_seq(p->b), p->qpos);
                q = bam_nt16_nt4_table[qq? qq : ref_base];
-               if (!p->is_del && q < 4) x |= 1 << 21 | q << 16, y |= q;
+               if (!p->is_del && !p->is_refskip && q < 4) x |= 1 << 21 | q << 16, y |= q;
                bm->aux->info16[n] = y;
                bm->aux->info[n++] = x;
        }
@@ -281,26 +281,25 @@ goto_glf:
 
 uint32_t glf2cns(const glf1_t *g, int q_r)
 {
-       int i, j, k, tmp[16], min = 10000, min2 = 10000, min3 = 10000, min_g = -1, min_g2 = -1;
+       int i, j, k, p[10], ref4;
        uint32_t x = 0;
+       ref4 = bam_nt16_nt4_table[g->ref_base];
        for (i = k = 0; i < 4; ++i)
                for (j = i; j < 4; ++j) {
-                       tmp[j<<2|i] = -1;
-                       tmp[i<<2|j] = g->lk[k++] + (i == j? 0 : q_r);
+                       int prior = (i == ref4 && j == ref4? 0 : i == ref4 || j == ref4? q_r : q_r + 3);
+                       p[k] = (g->lk[k] + prior)<<4 | i<<2 | j;
+                       ++k;
                }
-       for (i = 0; i < 16; ++i) {
-               if (tmp[i] < 0) continue;
-               if (tmp[i] < min) {
-                       min3 = min2; min2 = min; min = tmp[i]; min_g2 = min_g; min_g = i;
-               } else if (tmp[i] < min2) {
-                       min3 = min2; min2 = tmp[i]; min_g2 = i;
-               } else if (tmp[i] < min3) min3 = tmp[i];
-       }
-       x = min_g >= 0? (1U<<(min_g>>2&3) | 1U<<(min_g&3)) << 28 : 0xf << 28;
-       x |= min_g2 >= 0? (1U<<(min_g2>>2&3) | 1U<<(min_g2&3)) << 24 : 0xf << 24;
-       x |= (uint32_t)g->max_mapQ << 16;
-       x |= min2 < 10000? (min2 - min < 256? min2 - min : 255) << 8 : 0xff << 8;
-       x |= min2 < 10000 && min3 < 10000? (min3 - min2 < 256? min3 - min2 : 255) : 0xff;
+       for (i = 1; i < 10; ++i) // insertion sort
+               for (j = i; j > 0 && p[j] < p[j-1]; --j)
+                       k = p[j], p[j] = p[j-1], p[j-1] = k;
+       x = (1u<<(p[0]&3) | 1u<<(p[0]>>2&3)) << 28; // the best genotype
+       x |= (uint32_t)g->max_mapQ << 16; // rms mapQ
+       x |= ((p[1]>>4) - (p[0]>>4) < 256? (p[1]>>4) - (p[0]>>4) : 255) << 8; // consensus Q
+       for (k = 0; k < 10; ++k)
+               if ((p[k]&0xf) == (ref4<<2|ref4)) break;
+       if (k == 10) k = 9;
+       x |= (p[k]>>4) - (p[0]>>4) < 256? (p[k]>>4) - (p[0]>>4) : 255; // snp Q
        return x;
 }
 
@@ -310,7 +309,7 @@ uint32_t bam_maqcns_call(int n, const bam_pileup1_t *pl, bam_maqcns_t *bm)
        uint32_t x;
        if (n) {
                g = bam_maqcns_glfgen(n, pl, 0xf, bm);
-               x = glf2cns(g, (int)(bm->q_r + 0.5));
+               x = g->depth == 0? (0xfU<<28 | 0xfU<<24) : glf2cns(g, (int)(bm->q_r + 0.5));
                free(g);
        } else x = 0xfU<<28 | 0xfU<<24;
        return x;
@@ -461,7 +460,7 @@ bam_maqindel_ret_t *bam_maqindel(int n, int pos, const bam_maqindel_opt_t *mi, c
                for (i = 0; i < n_types; ++i) {
                        ka_param_t ap = ka_param_blast;
                        ap.band_width = 2 * types[n_types - 1] + 2;
-                       ap.gap_end = 0;
+                       ap.gap_end_ext = 0;
                        // write ref2
                        for (k = 0, j = left; j <= pos; ++j)
                                ref2[k++] = bam_nt16_nt4_table[bam_nt16_table[(int)ref[j]]];