X-Git-Url: https://git.donarmstrong.com/?a=blobdiff_plain;f=padding.c;h=6d97b1882a6f70ac6bca1d35f7b8217b0a02c926;hb=a46b631be3830db9bf4db53aa5e04c043a3213eb;hp=6e40833c9a7b20b24029b75f87b504f3503330d2;hpb=d2496f2aaae8022e04d10b9fd09eb305026a77df;p=samtools.git

diff --git a/padding.c b/padding.c
index 6e40833..6d97b18 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)
@@ -136,7 +149,7 @@ int bam_pad2unpad(bamFile in, bamFile out)
 					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) && (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 */
@@ -168,6 +181,7 @@ int main_pad2unpad(int argc, char *argv[])
         int result=0;
 	if (argc == 1) {
 		fprintf(stderr, "Usage: samtools depad <in.bam>\n");
+		fprintf(stderr, "\nRequires embedded reference sequences (before the reads for that reference).\n");
 		return 1;
 	}
 	in = strcmp(argv[1], "-")? bam_open(argv[1], "r") : bam_dopen(fileno(stdin), "r");