X-Git-Url: https://git.donarmstrong.com/?a=blobdiff_plain;f=padding.c;h=4061f247589dc641771b8b8da0bcba81475b49f3;hb=ccb7838cb53bf0ca6917a77f7991d940057c12db;hp=fe11da29e3bd285066fc3cb41cf80717c7b0cfc1;hpb=fc236a11b6b6ace1cb5ee572dded25a9dec07496;p=samtools.git

diff --git a/padding.c b/padding.c
index fe11da2..4061f24 100644
--- a/padding.c
+++ b/padding.c
@@ -30,9 +30,21 @@ static void replace_cigar(bam1_t *b, int n, uint32_t *cigar)
 static void unpad_seq(bam1_t *b, kstring_t *s)
 {
 	int k, j, i;
+	int length;
 	uint32_t *cigar = bam1_cigar(b);
 	uint8_t *seq = bam1_seq(b);
-	ks_resize(s, b->core.l_qseq);
+	// b->core.l_qseq gives length of the SEQ entry (including soft clips, S)
+	// We need the padded length after alignment from the CIGAR (excluding
+	// soft clips S, but including pads from CIGAR D operations)
+	length = 0;
+	for (k = 0; k < b->core.n_cigar; ++k) {
+		int op, ol;
+		op= bam_cigar_op(cigar[k]);
+		ol = bam_cigar_oplen(cigar[k]);
+		if (op == BAM_CMATCH || op == BAM_CEQUAL || op == BAM_CDIFF || op == BAM_CDEL)
+			length += ol;
+	}
+	ks_resize(s, length);
 	for (k = 0, s->l = 0, j = 0; k < b->core.n_cigar; ++k) {
 		int op, ol;
 		op = bam_cigar_op(cigar[k]);
@@ -43,13 +55,14 @@ static void unpad_seq(bam1_t *b, kstring_t *s)
 			j += ol;
 		} else if (op == BAM_CHARD_CLIP) {
 			/* do nothing */
-		} else if (op == BAM_CDEL || op == BAM_CPAD) {
+		} else if (op == BAM_CDEL) {
 			for (i = 0; i < ol; ++i) s->s[s->l++] = 0;
                 } else {
 			fprintf(stderr, "[depad] ERROR: Didn't expect CIGAR op %c in read %s\n", BAM_CIGAR_STR[op], bam1_qname(b));
                         assert(-1);
 		}
 	}
+	assert(length == s->l);
 }
 
 int bam_pad2unpad(bamFile in, bamFile out)
@@ -104,16 +117,15 @@ int bam_pad2unpad(bamFile in, bamFile out)
 				if (b->core.n_cigar > 2 && bam_cigar_op(cigar[1]) == BAM_CSOFT_CLIP) {
 					write_cigar(cigar2, n2, m2, cigar[1]);
 				}
-			} else {
-	  			/* Insert a dummy 0M entry to let use remove superfluous leading P ops */
-				write_cigar(cigar2, n2, m2, 0);
 			}
-			/* Include any pads if starts with an insert */
-			for (k = 0; k+1 < b->core.pos && !r.s[b->core.pos - k - 1]; ++k);
-			if (k) write_cigar(cigar2, n2, m2, bam_cigar_gen(k, BAM_CPAD));
 			/* Determine CIGAR operator for each base in the aligned read */
 			for (i = 0, k = b->core.pos; i < q.l; ++i, ++k)
 				q.s[i] = q.s[i]? (r.s[k]? BAM_CMATCH : BAM_CINS) : (r.s[k]? BAM_CDEL : BAM_CPAD);
+			/* Include any pads if starts with an insert */
+			if (q.s[0] == BAM_CINS) {
+				for (k = 0; k+1 < b->core.pos && !r.s[b->core.pos - k - 1]; ++k);
+				if (k) write_cigar(cigar2, n2, m2, bam_cigar_gen(k, BAM_CPAD));
+			}
 			/* Count consecutive CIGAR operators to turn into a CIGAR string */
 			for (i = k = 1, op = q.s[0]; i < q.l; ++i) {
 				if (op != q.s[i]) {
@@ -131,14 +143,13 @@ int bam_pad2unpad(bamFile in, bamFile out)
 				write_cigar(cigar2, n2, m2, cigar[b->core.n_cigar-1]);
 			}
 			/* Remove redundant P operators between M/X/=/D operators, e.g. 5M2P10M -> 15M */
-			/* or right at the start before M/X/=/D */
 			int pre_op, post_op;
 			for (i = 2; i < n2; ++i)
 				if (bam_cigar_op(cigar2[i-1]) == BAM_CPAD) {
 					pre_op = bam_cigar_op(cigar2[i-2]);
 					post_op = bam_cigar_op(cigar2[i]);
 					/* Note don't need to check for X/= as code above will use M only */
-					if ((pre_op == BAM_CMATCH || pre_op == BAM_CDIFF || pre_op == BAM_CSOFT_CLIP || pre_op == BAM_CHARD_CLIP) || (post_op == BAM_CMATCH || post_op == BAM_CDIFF)) {
+					if ((pre_op == BAM_CMATCH || pre_op == BAM_CDEL) && (post_op == BAM_CMATCH || post_op == BAM_CDEL)) {
 						/* This is a redundant P operator */
 						cigar2[i-1] = 0; // i.e. 0M
 						/* If had same operator either side, combine them in post_op */