bamcheck: new stats, number of filtered vs raw sequences

[samtools.git] / bam2bcf_indel.c

diff --git a/bam2bcf_indel.c b/bam2bcf_indel.c
index f37a7831f9625be14b416510374fd5f71150cfdd..9aea5f9d669f79850fb45438fdcf1183c3e1227f 100644 (file)
--- a/bam2bcf_indel.c
+++ b/bam2bcf_indel.c
@@ -66,7 +66,7 @@ static int tpos2qpos(const bam1_core_t *c, const uint32_t *cigar, int32_t tpos,
        for (k = 0; k < c->n_cigar; ++k) {
                int op = cigar[k] & BAM_CIGAR_MASK;
                int l = cigar[k] >> BAM_CIGAR_SHIFT;
-               if (op == BAM_CMATCH) {
+               if (op == BAM_CMATCH || op == BAM_CEQUAL || op == BAM_CDIFF) {
                        if (c->pos > tpos) return y;
                        if (x + l > tpos) {
                                *_tpos = tpos;
@@ -142,31 +142,53 @@ int bcf_call_gap_prep(int n, int *n_plp, bam_pileup1_t **plp, int pos, bcf_calla
        if (s == n) return -1; // there is no indel at this position.
        for (s = N = 0; s < n; ++s) N += n_plp[s]; // N is the total number of reads
        { // find out how many types of indels are present
-               int m, n_alt = 0, n_tot = 0;
+        bca->max_support = bca->max_frac = 0;
+               int m, n_alt = 0, n_tot = 0, indel_support_ok = 0;
                uint32_t *aux;
                aux = calloc(N + 1, 4);
                m = max_rd_len = 0;
                aux[m++] = MINUS_CONST; // zero indel is always a type
                for (s = 0; s < n; ++s) {
+            int na = 0, nt = 0; 
                        for (i = 0; i < n_plp[s]; ++i) {
                                const bam_pileup1_t *p = plp[s] + i;
                                if (rghash == 0 || p->aux == 0) {
-                                       ++n_tot;
+                                       ++nt;
                                        if (p->indel != 0) {
-                                               ++n_alt;
+                                               ++na;
                                                aux[m++] = MINUS_CONST + p->indel;
                                        }
                                }
                                j = bam_cigar2qlen(&p->b->core, bam1_cigar(p->b));
                                if (j > max_rd_len) max_rd_len = j;
                        }
+            float frac = (float)na/nt;
+            if ( !indel_support_ok && na >= bca->min_support && frac >= bca->min_frac )
+                indel_support_ok = 1;
+            if ( na > bca->max_support && frac > 0 ) bca->max_support = na, bca->max_frac = frac;
+            n_alt += na; 
+            n_tot += nt;
                }
+        // To prevent long stretches of N's to be mistaken for indels (sometimes thousands of bases),
+        //  check the number of N's in the sequence and skip places where half or more reference bases are Ns.
+        int nN=0; for (i=pos; i-pos<max_rd_len && ref[i]; i++) if ( ref[i]=='N' ) nN++;
+        if ( nN*2>i ) { free(aux); return -1; }
+
                ks_introsort(uint32_t, m, aux);
                // squeeze out identical types
                for (i = 1, n_types = 1; i < m; ++i)
                        if (aux[i] != aux[i-1]) ++n_types;
-               if (n_types == 1 || (double)n_alt / n_tot < bca->min_frac || n_alt < bca->min_support) { // then skip
-                       free(aux); return -1;
+        // Taking totals makes it hard to call rare indels
+        if ( !bca->per_sample_flt )
+            indel_support_ok = ( (float)n_alt / n_tot < bca->min_frac || n_alt < bca->min_support ) ? 0 : 1;
+        if ( n_types == 1 || !indel_support_ok ) { // then skip
+            free(aux); return -1;
+        }
+               if (n_types >= 64) {
+                       free(aux);
+                       if (bam_verbose >= 2) 
+                               fprintf(stderr, "[%s] excessive INDEL alleles at position %d. Skip the position.\n", __func__, pos + 1);
+                       return -1;
                }
                types = (int*)calloc(n_types, sizeof(int));
                t = 0;
@@ -178,7 +200,6 @@ int bcf_call_gap_prep(int n, int *n_plp, bam_pileup1_t **plp, int pos, bcf_calla
                for (t = 0; t < n_types; ++t)
                        if (types[t] == 0) break;
                ref_type = t; // the index of the reference type (0)
-               assert(n_types < 64);
        }
        { // calculate left and right boundary
                left = pos > INDEL_WINDOW_SIZE? pos - INDEL_WINDOW_SIZE : 0;
@@ -217,7 +238,7 @@ int bcf_call_gap_prep(int n, int *n_plp, bam_pileup1_t **plp, int pos, bcf_calla
                                for (k = 0; k < b->core.n_cigar; ++k) {
                                        int op = cigar[k]&0xf;
                                        int j, l = cigar[k]>>4;
-                                       if (op == BAM_CMATCH) {
+                                       if (op == BAM_CMATCH || op == BAM_CEQUAL || op == BAM_CDIFF) {
                                                for (j = 0; j < l; ++j)
                                                        if (x + j >= left && x + j < right)
                                                                cns[x+j-left] += (bam1_seqi(seq, y+j) == ref0[x+j-left])? 1 : 0x10000;