In Casava 1.8 the fastq output changed, the name had a space which bwa

wasn't parsing correctly. This patch fixes that and enables bwa to filter
sequences marked by Casava, removing this tag from the output.

Signed-off-by: RoelKluin <roel.kluin@gmail.com>

bwaseqio.c

@@ -157,6 +157,25 @@ bwa_seq_t *bwa_read_seq(bwa_seqio_t *bs, int n_needed, int *n, int mode, int tri
 	n_seqs = 0;
 	seqs = (bwa_seq_t*)calloc(n_needed, sizeof(bwa_seq_t));
 	while ((l = kseq_read(seq)) >= 0) {
+		// skip reads that are marked to be filtered by Casava
+		if (mode & BWA_MODE_CFY) {
+			char *s = rindex(seq->name.s, ' ');
+			if (s) {
+				*s = '\0';
+				for(++s; *s != '\0'; ++s) {
+					if (*s == ':') {
+						++s;
+						break;
+					}
+				}
+				if (*s == 'Y')
+					continue;
+			}
+			if (!s || *s != 'N') {
+				fprintf(stderr, "No Casava filter character found.\n");
+				return 0;
+			}
+		}
 		if (is_64 && seq->qual.l)
 			for (i = 0; i < seq->qual.l; ++i) seq->qual.s[i] -= 31;
 		if (seq->seq.l <= l_bc) continue; // sequence length equals or smaller than the barcode length

bwtaln.c

@@ -233,7 +233,7 @@ int bwa_aln(int argc, char *argv[])
 	gap_opt_t *opt;
 
 	opt = gap_init_opt();
-	while ((c = getopt(argc, argv, "n:o:e:i:d:l:k:cLR:m:t:NM:O:E:q:f:b012IB:")) >= 0) {
+	while ((c = getopt(argc, argv, "n:o:e:i:d:l:k:cLR:m:t:NM:O:E:q:f:b012IYB:")) >= 0) {
 		switch (c) {
 		case 'n':
 			if (strstr(optarg, ".")) opt->fnr = atof(optarg), opt->max_diff = -1;
@@ -261,6 +261,7 @@ int bwa_aln(int argc, char *argv[])
 		case '1': opt->mode |= BWA_MODE_BAM_READ1; break;
 		case '2': opt->mode |= BWA_MODE_BAM_READ2; break;
 		case 'I': opt->mode |= BWA_MODE_IL13; break;
+		case 'Y': opt->mode |= BWA_MODE_CFY; break;
 		case 'B': opt->mode |= atoi(optarg) << 24; break;
 		default: return 1;
 		}
@@ -298,6 +299,7 @@ int bwa_aln(int argc, char *argv[])
 		fprintf(stderr, "         -0        use single-end reads only (effective with -b)\n");
 		fprintf(stderr, "         -1        use the 1st read in a pair (effective with -b)\n");
 		fprintf(stderr, "         -2        use the 2nd read in a pair (effective with -b)\n");
+		fprintf(stderr, "         -Y        filter Casava-filtered sequences\n");
 		fprintf(stderr, "\n");
 		return 1;
 	}

bwtaln.h

@@ -86,6 +86,7 @@ typedef struct {
 #define BWA_MODE_GAPE       0x01
 #define BWA_MODE_COMPREAD   0x02
 #define BWA_MODE_LOGGAP     0x04
+#define BWA_MODE_CFY        0x08
 #define BWA_MODE_NONSTOP    0x10
 #define BWA_MODE_BAM        0x20
 #define BWA_MODE_BAM_SE     0x40

kseq.h

@@ -102,7 +102,7 @@ typedef struct __kstring_t {
 					if (ks->buf[i] == delimiter) break;					\
 			} else {													\
 				for (i = ks->begin; i < ks->end; ++i)					\
-					if (isspace(ks->buf[i])) break;						\
+					if (isspace(ks->buf[i]) && (ks->buf[i] != ' ')) break;						\
 			}															\
 			if (str->m - str->l < i - ks->begin + 1) {					\
 				str->m = str->l + (i - ks->begin) + 1;					\