* samtools-0.1.9-9 (r810)

[samtools.git] / bam_md.c

diff --git a/bam_md.c b/bam_md.c
index f65b845b5d594980d4b4dba36122672c71b19cc9..a8ddd1baf8aa78d6911ca7a804a549a1d27eb06b 100644 (file)
--- a/bam_md.c
+++ b/bam_md.c
@@ -7,6 +7,7 @@
 #include "sam.h"
 #include "kstring.h"
 #include "kaln.h"
+#include "kprobaln.h"
 
 void bam_fillmd1_core(bam1_t *b, char *ref, int is_equal, int max_nm)
 {
@@ -161,12 +162,12 @@ int bam_cap_mapQ(bam1_t *b, char *ref, int thres)
        return (int)(t + .499);
 }
 
-int bam_prob_realn(bam1_t *b, const char *ref)
+int bam_prob_realn_core(bam1_t *b, const char *ref, int write_bq)
 {
        int k, i, bw, x, y, yb, ye, xb, xe;
        uint32_t *cigar = bam1_cigar(b);
        bam1_core_t *c = &b->core;
-       ka_probpar_t conf = ka_probpar_def;
+       kpa_par_t conf = kpa_par_def;
        // find the start and end of the alignment
        if (c->flag & BAM_FUNMAP) return -1;
        x = c->pos, y = 0, yb = ye = xb = xe = -1;
@@ -192,8 +193,12 @@ int bam_prob_realn(bam1_t *b, const char *ref)
        if (xe - xb - c->l_qseq > bw)
                xb += (xe - xb - c->l_qseq - bw) / 2, xe -= (xe - xb - c->l_qseq - bw) / 2;
        { // glocal
-               uint8_t *s, *r, *q, *seq = bam1_seq(b), *qual = bam1_qual(b);
+               uint8_t *s, *r, *q, *seq = bam1_seq(b), *qual = bam1_qual(b), *bq = 0;
                int *state;
+               if (write_bq) {
+                       bq = calloc(c->l_qseq + 1, 1);
+                       memcpy(bq, qual, c->l_qseq);
+               }
                s = calloc(c->l_qseq, 1);
                for (i = 0; i < c->l_qseq; ++i) s[i] = bam_nt16_nt4_table[bam1_seqi(seq, i)];
                r = calloc(xe - xb, 1);
@@ -201,7 +206,7 @@ int bam_prob_realn(bam1_t *b, const char *ref)
                        r[i-xb] = bam_nt16_nt4_table[bam_nt16_table[(int)ref[i]]];
                state = calloc(c->l_qseq, sizeof(int));
                q = calloc(c->l_qseq, 1);
-               ka_prob_glocal(r, xe-xb, s, c->l_qseq, qual, &conf, state, q);
+               kpa_glocal(r, xe-xb, s, c->l_qseq, qual, &conf, state, q);
                for (k = 0, x = c->pos, y = 0; k < c->n_cigar; ++k) {
                        int op = cigar[k]&0xf, l = cigar[k]>>4;
                        if (op == BAM_CMATCH) {
@@ -213,11 +218,21 @@ int bam_prob_realn(bam1_t *b, const char *ref)
                        } else if (op == BAM_CSOFT_CLIP || op == BAM_CINS) y += l;
                        else if (op == BAM_CDEL) x += l;
                }
+               if (write_bq) {
+                       for (i = 0; i < c->l_qseq; ++i) bq[i] = bq[i] - qual[i] + 33;
+                       bam_aux_append(b, "BQ", 'Z', c->l_qseq + 1, bq);
+                       free(bq);
+               }
                free(s); free(r); free(q); free(state);
        }
        return 0;
 }
 
+int bam_prob_realn(bam1_t *b, const char *ref)
+{
+       return bam_prob_realn_core(b, ref, 0);
+}
+
 int bam_fillmd(int argc, char *argv[])
 {
        int c, is_equal = 0, tid = -2, ret, len, is_bam_out, is_sam_in, is_uncompressed, max_nm = 0, is_realn, capQ = 0;
@@ -275,8 +290,7 @@ int bam_fillmd(int argc, char *argv[])
                                        fprintf(stderr, "[bam_fillmd] fail to find sequence '%s' in the reference.\n",
                                                        fp->header->target_name[tid]);
                        }
-//                     if (is_realn) bam_realn(b, ref);
-                       if (is_realn) bam_prob_realn(b, ref);
+                       if (is_realn) bam_prob_realn_core(b, ref, 0);
                        if (capQ > 10) {
                                int q = bam_cap_mapQ(b, ref, capQ);
                                if (b->core.qual > q) b->core.qual = q;