Added error detection for cases such as a read's two mates having different names...

[rsem.git] / SamParser.h

diff --git a/SamParser.h b/SamParser.h
index c5a503554b8c9cf4716b722d47da5f8c5db8540c..4747ef4281af593f714450d68ccfc65090662891 100644 (file)
--- a/SamParser.h
+++ b/SamParser.h
@@ -25,7 +25,7 @@
 
 class SamParser {
 public:
-       SamParser(char, const char*, Transcripts&, const char* = 0);
+       SamParser(char, const char*, const char*, Transcripts&, const char*);
        ~SamParser();
 
        /**
@@ -62,7 +62,7 @@ private:
        }
 
        std::string getName(const bam1_t* b) {
-               return std::string((char*)bam1_qname(b));
+               return std::string(bam1_qname(b));
        }
 
        std::string getReadSeq(const bam1_t*);
@@ -80,7 +80,7 @@ private:
 char SamParser::rtTag[STRLEN] = ""; // default : no tag, thus no Type 2 reads
 
 // aux, if not 0, points to the file name of fn_list
-SamParser::SamParser(char inpType, const char* inpF, Transcripts& transcripts, const char* aux)
+SamParser::SamParser(char inpType, const char* inpF, const char* aux, Transcripts& transcripts, const char* imdName)
        : transcripts(transcripts)
 {
        switch(inpType) {
@@ -93,7 +93,7 @@ SamParser::SamParser(char inpType, const char* inpF, Transcripts& transcripts, c
        header = sam_in->header;
        general_assert(header != 0, "Fail to parse sam header!");
 
-       transcripts.buildMappings(header->n_targets, header->target_name);
+       transcripts.buildMappings(header->n_targets, header->target_name, imdName);
 
        b = bam_init1();
        b2 = bam_init1();
@@ -112,8 +112,7 @@ int SamParser::parseNext(SingleRead& read, SingleHit& hit) {
        bool canR = (samread(sam_in, b) >= 0);
        if (!canR) return -1;
 
-       if (b->core.flag & 0x0001) { fprintf(stderr, "Find a paired end read in the file!\n"); exit(-1); }
-       //(b->core.flag & 0x0100) &&  && !(b->core.flag & 0x0004)
+       general_assert(!(b->core.flag & 0x0001), "Find a paired end read in the file!");
 
        int readType = getReadType(b);
        std::string name = getName(b);
@@ -143,8 +142,7 @@ int SamParser::parseNext(SingleReadQ& read, SingleHit& hit) {
        bool canR = (samread(sam_in, b) >= 0);
        if (!canR) return -1;
 
-       if (b->core.flag & 0x0001) { fprintf(stderr, "Find a paired end read in the file!\n"); exit(-1); }
-       //assert(!(b->core.flag & 0x0001)); //(b->core.flag & 0x0100) &&  && !(b->core.flag & 0x0004)
+       general_assert(!(b->core.flag & 0x0001), "Find a paired end read in the file!");
 
        int readType = getReadType(b);
        std::string name = getName(b);
@@ -175,21 +173,24 @@ int SamParser::parseNext(PairedEndRead& read, PairedEndHit& hit) {
        bool canR = ((samread(sam_in, b) >= 0) && (samread(sam_in, b2) >= 0));
        if (!canR) return -1;
 
-       if (!((b->core.flag & 0x0001) && (b2->core.flag & 0x0001))) {
-               fprintf(stderr, "One of the mate is not paired-end! (RSEM assumes the two mates of a paired-end read should be adjacent)\n");
-               exit(-1);
-       }
-       //assert((b->core.flag & 0x0001) && (b2->core.flag & 0x0001));
+       general_assert((b->core.flag & 0x0001) && (b2->core.flag & 0x0001), \
+                       "One of the mate is not paired-end! (RSEM assumes the two mates of a paired-end read should be adjacent)");
 
        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?
+
+       general_assert(!strcmp(bam1_qname(mp1), bam1_qname(mp2)), "Detected a read pair whose two mates have different names: " + getName(mp1) + " , " + getName(mp2) + " !");
 
        int readType = getReadType(mp1, mp2);
        std::string name = getName(mp1);
@@ -226,21 +227,24 @@ int SamParser::parseNext(PairedEndReadQ& read, PairedEndHit& hit) {
        bool canR = ((samread(sam_in, b) >= 0) && (samread(sam_in, b2) >= 0));
        if (!canR) return -1;
 
-       if (!((b->core.flag & 0x0001) && (b2->core.flag & 0x0001))) {
-               fprintf(stderr, "One of the mate is not paired-end! (RSEM assumes the two mates of a paired-end read should be adjacent)\n");
-               exit(-1);
-       }
-       //assert((b->core.flag & 0x0001) && (b2->core.flag & 0x0001));
+       general_assert((b->core.flag & 0x0001) && (b2->core.flag & 0x0001), \
+                       "One of the mate is not paired-end! (RSEM assumes the two mates of a paired-end read should be adjacent)");
 
        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;
+
+       general_assert(!strcmp(bam1_qname(mp1), bam1_qname(mp2)), "Detected a read pair whose two mates have different names: " + getName(mp1) + " , " + getName(mp2) + " !");
 
        int readType = getReadType(mp1, mp2);
        std::string name = getName(mp1);
@@ -341,10 +345,17 @@ 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;
 }