For genome BAM, modified MD tag accordingly

[rsem.git] / SamParser.h

diff --git a/SamParser.h b/SamParser.h
index 9d62eff0402ba2eca240b686b598ea3b36324ea4..e91f9c379539fc7ada0f8682bcfc7c957489d67f 100644 (file)
--- a/SamParser.h
+++ b/SamParser.h
@@ -178,13 +178,17 @@ int SamParser::parseNext(PairedEndRead& read, PairedEndHit& hit) {
 
        bam1_t *mp1 = NULL, *mp2 = NULL;
 
-       if ((b->core.flag & 0x0040) && (b2->core.flag & 0x0080)) {
+       // If lose mate info, discard. is it necessary?
+       if (!((b->core.flag & 0x0040) && (b2->core.flag & 0x0080)) && !((b->core.flag & 0x0080) && (b2->core.flag & 0x0040))) return 4;
+       // If only one mate is mapped, discard
+       if (((b->core.flag & 0x0004) && !(b2->core.flag & 0x0004)) || (!(b->core.flag & 0x0004) && (b2->core.flag & 0x0004))) return 4;
+
+       if (b->core.flag & 0x0040) {
                mp1 = b; mp2 = b2;
        }
-       else if ((b->core.flag & 0x0080) && (b2->core.flag & 0x0040)) {
+       else  {
                mp1 = b2; mp2 = b;
        }
-       else return 4; // If lose mate info, discard. is it necessary?
 
        int readType = getReadType(mp1, mp2);
        std::string name = getName(mp1);
@@ -226,13 +230,17 @@ int SamParser::parseNext(PairedEndReadQ& read, PairedEndHit& hit) {
 
        bam1_t *mp1 = NULL, *mp2 = NULL;
 
-       if ((b->core.flag & 0x0040) && (b2->core.flag & 0x0080)) {
+       // If lose mate info, discard. is it necessary?
+       if (!((b->core.flag & 0x0040) && (b2->core.flag & 0x0080)) && !((b->core.flag & 0x0080) && (b2->core.flag & 0x0040))) return 4;
+       // If only one mate is mapped, discard
+       if (((b->core.flag & 0x0004) && !(b2->core.flag & 0x0004)) || (!(b->core.flag & 0x0004) && (b2->core.flag & 0x0004))) return 4;
+
+       if (b->core.flag & 0x0040) {
                mp1 = b; mp2 = b2;
        }
-       else if ((b->core.flag & 0x0080) && (b2->core.flag & 0x0040)) {
+       else  {
                mp1 = b2; mp2 = b;
        }
-       else return 4;
 
        int readType = getReadType(mp1, mp2);
        std::string name = getName(mp1);
@@ -333,10 +341,17 @@ inline int SamParser::getReadType(const bam1_t* b) {
        return (bam_aux2i(p) > 0 ? 2 : 0);
 }
 
-
 //For paired-end reads, do not print out type 2 reads
 inline int SamParser::getReadType(const bam1_t* b, const bam1_t* b2) {
-       if ((b->core.flag & 0x0002) && (b2->core.flag & 0x0002)) return 1;
+       if (!(b->core.flag & 0x0004) && !(b2->core.flag & 0x0004)) return 1;
+
+       if (!strcmp(rtTag, "")) return 0;
+
+       uint8_t *p = bam_aux_get(b, rtTag);
+       if (p != NULL && bam_aux2i(p) > 0) return 2;
+
+       p = bam_aux_get(b2, rtTag);
+       if (p != NULL && bam_aux2i(p) > 0) return 2;
 
        return 0;
 }