X-Git-Url: https://git.donarmstrong.com/?a=blobdiff_plain;f=bam2bcf_indel.c;h=ab9e83ca8f712ae7d7975d67b94b10c702e64908;hb=ccb7838cb53bf0ca6917a77f7991d940057c12db;hp=f37a7831f9625be14b416510374fd5f71150cfdd;hpb=c350a570e955d5b4c2c13f607b7442df6d332c67;p=samtools.git

diff --git a/bam2bcf_indel.c b/bam2bcf_indel.c
index f37a783..ab9e83c 100644
--- 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;
@@ -161,6 +161,11 @@ int bcf_call_gap_prep(int n, int *n_plp, bam_pileup1_t **plp, int pos, bcf_calla
 				if (j > max_rd_len) max_rd_len = j;
 			}
 		}
+        // 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 ) return -1;
+
 		ks_introsort(uint32_t, m, aux);
 		// squeeze out identical types
 		for (i = 1, n_types = 1; i < m; ++i)
@@ -168,6 +173,12 @@ int bcf_call_gap_prep(int n, int *n_plp, bam_pileup1_t **plp, int pos, bcf_calla
 		if (n_types == 1 || (double)n_alt / n_tot < bca->min_frac || n_alt < bca->min_support) { // 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;
 		types[t++] = aux[0] - MINUS_CONST; 
@@ -178,7 +189,6 @@ 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 +227,7 @@ 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;