Modified the acknowledgement section of README.md

[rsem.git] / SamParser.h

diff --git a/SamParser.h b/SamParser.h
index a36af3a0d11a7133b27c5ee4fe83ae7996b9f631..e91f9c379539fc7ada0f8682bcfc7c957489d67f 100644 (file)
--- a/SamParser.h
+++ b/SamParser.h
@@ -1,4 +1,4 @@
-/* ReadType here means if the read is unalignable, alignable or aligned too much. It is NOT single read or paired-end read */
+/* ReadType here means if the read is unalignable, alignable or aligned too much. It is NOT siheaderngle read or paired-end read */
 #ifndef SAMPARSER_H_
 #define SAMPARSER_H_
 
@@ -10,7 +10,10 @@
 
 #include "sam/bam.h"
 #include "sam/sam.h"
+
 #include "utils.h"
+#include "my_assert.h"
+
 #include "SingleRead.h"
 #include "SingleReadQ.h"
 #include "PairedEndRead.h"
@@ -18,9 +21,11 @@
 #include "SingleHit.h"
 #include "PairedEndHit.h"
 
+#include "Transcripts.h"
+
 class SamParser {
 public:
-       SamParser(char, const char*, const char* = 0);
+       SamParser(char, const char*, Transcripts&, const char* = 0);
        ~SamParser();
 
        /**
@@ -46,6 +51,8 @@ private:
        bam_header_t *header;
        bam1_t *b, *b2;
 
+       Transcripts& transcripts;
+
        //tag used by aligner
        static char rtTag[STRLEN];
 
@@ -64,24 +71,32 @@ private:
        //0 ~ N0 1 ~ N1 2 ~ N2
        int getReadType(const bam1_t*);
        int getReadType(const bam1_t*, const bam1_t*); // for paired-end reads
+
+       bool check(bam1_t *b) {
+               return (b->core.n_cigar == 1) && ((*bam1_cigar(b) & BAM_CIGAR_MASK) == BAM_CMATCH) && (b->core.l_qseq == (int32_t)(*bam1_cigar(b) >> BAM_CIGAR_SHIFT));
+       }
 };
 
 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, const char* aux) {
+SamParser::SamParser(char inpType, const char* inpF, Transcripts& transcripts, const char* aux)
+       : transcripts(transcripts)
+{
        switch(inpType) {
        case 'b': sam_in = samopen(inpF, "rb", aux); break;
        case 's': sam_in = samopen(inpF, "r", aux); break;
        default: assert(false);
        }
 
-       if (sam_in == 0) { fprintf(stderr, "Cannot open %s! It may not exist.\n", inpF); exit(-1); }
-    header = sam_in->header;
-    if (header == 0) { fprintf(stderr, "Fail to parse sam header!\n"); exit(-1); }
+       general_assert(sam_in != 0, "Cannot open " + cstrtos(inpF) + "! It may not exist.");
+       header = sam_in->header;
+       general_assert(header != 0, "Fail to parse sam header!");
 
-    b = bam_init1();
-    b2 = bam_init1();
+       transcripts.buildMappings(header->n_targets, header->target_name);
+
+       b = bam_init1();
+       b2 = bam_init1();
 }
 
 SamParser::~SamParser() {
@@ -97,8 +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);
@@ -110,11 +124,13 @@ int SamParser::parseNext(SingleRead& read, SingleHit& hit) {
        else val = 5;
 
        if (readType == 1) {
+               if (!check(b)) { fprintf(stderr, "RSEM does not support gapped alignments, sorry!\n"); exit(-1); }
+
                if (getDir(b) > 0) {
-                       hit = SingleHit(b->core.tid + 1, b->core.pos);
+                       hit = SingleHit(transcripts.getInternalSid(b->core.tid + 1), b->core.pos);
                }
                else {
-                       hit = SingleHit(-(b->core.tid + 1), header->target_len[b->core.tid] - b->core.pos - b->core.l_qseq);
+                       hit = SingleHit(-transcripts.getInternalSid(b->core.tid + 1), header->target_len[b->core.tid] - b->core.pos - b->core.l_qseq);
                }
        }
 
@@ -126,8 +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);
@@ -139,11 +154,13 @@ int SamParser::parseNext(SingleReadQ& read, SingleHit& hit) {
        else val = 5;
 
        if (readType == 1) {
+               if (!check(b)) { fprintf(stderr, "RSEM does not support gapped alignments, sorry!\n"); exit(-1); }
+
                if (getDir(b) > 0) {
-                       hit = SingleHit(b->core.tid + 1, b->core.pos);
+                       hit = SingleHit(transcripts.getInternalSid(b->core.tid + 1), b->core.pos);
                }
                else {
-                       hit = SingleHit(-(b->core.tid + 1), header->target_len[b->core.tid] - b->core.pos - b->core.l_qseq);
+                       hit = SingleHit(-transcripts.getInternalSid(b->core.tid + 1), header->target_len[b->core.tid] - b->core.pos - b->core.l_qseq);
                }
        }
 
@@ -156,21 +173,22 @@ 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?
 
        int readType = getReadType(mp1, mp2);
        std::string name = getName(mp1);
@@ -184,16 +202,18 @@ int SamParser::parseNext(PairedEndRead& read, PairedEndHit& hit) {
        else val = 5;
 
        if (readType == 1) {
+               if (!check(mp1) || !check(mp2)) { fprintf(stderr, "RSEM does not support gapped alignments, sorry!\n"); exit(-1); }
+
                if (mp1->core.tid != mp2->core.tid) {
                        fprintf(stderr, "The two reads do not come from the same pair!");
                        exit(-1);
                }
                //assert(mp1->core.tid == mp2->core.tid);
                if (getDir(mp1) > 0) {
-                       hit = PairedEndHit(mp1->core.tid + 1, mp1->core.pos, mp2->core.pos + mp2->core.l_qseq - mp1->core.pos);
+                       hit = PairedEndHit(transcripts.getInternalSid(mp1->core.tid + 1), mp1->core.pos, mp2->core.pos + mp2->core.l_qseq - mp1->core.pos);
                }
                else {
-                       hit = PairedEndHit(-(mp1->core.tid + 1), header->target_len[mp1->core.tid] - mp1->core.pos - mp1->core.l_qseq, mp1->core.pos + mp1->core.l_qseq - mp2->core.pos);
+                       hit = PairedEndHit(-transcripts.getInternalSid(mp1->core.tid + 1), header->target_len[mp1->core.tid] - mp1->core.pos - mp1->core.l_qseq, mp1->core.pos + mp1->core.l_qseq - mp2->core.pos);
                }
        }
 
@@ -205,21 +225,22 @@ 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;
 
        int readType = getReadType(mp1, mp2);
        std::string name = getName(mp1);
@@ -233,16 +254,18 @@ int SamParser::parseNext(PairedEndReadQ& read, PairedEndHit& hit) {
        else val = 5;
 
        if (readType == 1) {
+               if (!check(mp1) || !check(mp2)) { fprintf(stderr, "RSEM does not support gapped alignments, sorry!\n"); exit(-1); }
+
                if (mp1->core.tid != mp2->core.tid) {
                        fprintf(stderr, "The two reads do not come from the same pair!");
                        exit(-1);
                }
                //assert(mp1->core.tid == mp2->core.tid);
                if (getDir(mp1) > 0) {
-                       hit = PairedEndHit(mp1->core.tid + 1, mp1->core.pos, mp2->core.pos + mp2->core.l_qseq - mp1->core.pos);
+                       hit = PairedEndHit(transcripts.getInternalSid(mp1->core.tid + 1), mp1->core.pos, mp2->core.pos + mp2->core.l_qseq - mp1->core.pos);
                }
                else {
-                       hit = PairedEndHit(-(mp1->core.tid + 1), header->target_len[mp1->core.tid] - mp1->core.pos - mp1->core.l_qseq, mp1->core.pos + mp1->core.l_qseq - mp2->core.pos);
+                       hit = PairedEndHit(-transcripts.getInternalSid(mp1->core.tid + 1), header->target_len[mp1->core.tid] - mp1->core.pos - mp1->core.l_qseq, mp1->core.pos + mp1->core.l_qseq - mp2->core.pos);
                }
        }
 
@@ -318,10 +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;
 }