Fixes 1-based error for anti-sense reads and TLEN in pseudobam

pachterlab · Jun 17, 2015 · 4dce074 · 4dce074
1 parent a2ee0b2
commit 4dce074
Show file tree

Hide file tree

Showing 2 changed files with 22 additions and 10 deletions.
diff --git a/src/KmerIndex.cpp b/src/KmerIndex.cpp
@@ -951,6 +951,7 @@ void KmerIndex::load(ProgramOptions& opt, bool loadKmerTable) {
   in.close();
 }
 
+
 int KmerIndex::mapPair(const char *s1, int l1, const char *s2, int l2, int ec) const {
   bool d1 = true;
   bool d2 = true;
@@ -1171,6 +1172,12 @@ void KmerIndex::match(const char *s, int l, std::vector<std::pair<KmerEntry, int
   }
 }
 
+
+//use:  (pos,sense) = index.findPosition(tr,km,val,p)
+//pre:  index.kmap[km] == val,
+//      km is the p-th k-mer of a read
+//      val.contig maps to tr
+//post: km is found in position pos (1-based) on the sense/!sense strand of tr
 std::pair<int,bool> KmerIndex::findPosition(int tr, Kmer km, KmerEntry val, int p) const {
   bool fw = (km == km.rep());
   bool csense = (fw == val.isFw());
@@ -1207,7 +1214,6 @@ std::pair<int,bool> KmerIndex::findPosition(int tr, Kmer km, KmerEntry val, int
       return {trpos + (c.length - val.getPos())  - p, !csense}; // 1-based, case II
     }
   }
-
 }
 
 /*

diff --git a/src/ProcessReads.cpp b/src/ProcessReads.cpp
@@ -395,13 +395,16 @@ void outputPseudoBam(const KmerIndex &index, int ec, const kseq_t *seq1, const s
           }
         }
 
-        int posread = (f1 & 0x10) ? (x1.first - seq1->seq.l) : x1.first;
-        int posmate = (f1 & 0x20) ? (x2.first - seq2->seq.l) : x2.first;
+        int posread = (f1 & 0x10) ? (x1.first - seq1->seq.l + 1) : x1.first;
+        int posmate = (f1 & 0x20) ? (x2.first - seq2->seq.l + 1) : x2.first;
 
         getCIGARandSoftClip(cig, bool(f1 & 0x10), (f1 & 0x04) == 0, posread, posmate, seq1->seq.l, index.target_lens_[tr]);
-
+        int tlen = x2.first - x1.first;
+        if (tlen != 0) {
+          tlen += (tlen>0) ? 1 : -1;
+        }
 
-        printf("%s\t%d\t%s\t%d\t255\t%s\t=\t%d\t%d\t%s\t%s\tNH:i:%d\n", seq1->name.s, f1 & 0xFFFF, index.target_names_[tr].c_str(), posread, cig, posmate, (x2.first-x1.first), (f1 & 0x10) ? &buf1[0] : seq1->seq.s, (f1 & 0x10) ? &buf2[0] : seq1->qual.s, nmap);
+        printf("%s\t%d\t%s\t%d\t255\t%s\t=\t%d\t%d\t%s\t%s\tNH:i:%d\n", seq1->name.s, f1 & 0xFFFF, index.target_names_[tr].c_str(), posread, cig, posmate, tlen, (f1 & 0x10) ? &buf1[0] : seq1->seq.s, (f1 & 0x10) ? &buf2[0] : seq1->qual.s, nmap);
       }
 
       revset = false;
@@ -434,13 +437,16 @@ void outputPseudoBam(const KmerIndex &index, int ec, const kseq_t *seq1, const s
           }
         }
 
-        int posread = (f2 & 0x10) ? (x2.first - seq2->seq.l) : x2.first;
-        int posmate = (f2 & 0x20) ? (x1.first - seq1->seq.l) : x1.first;
-
+        int posread = (f2 & 0x10) ? (x2.first - seq2->seq.l + 1) : x2.first;
+        int posmate = (f2 & 0x20) ? (x1.first - seq1->seq.l + 1) : x1.first;
 
         getCIGARandSoftClip(cig, bool(f2 & 0x10), (f2 & 0x04) == 0, posread, posmate, seq2->seq.l, index.target_lens_[tr]);
+        int tlen = x1.first - x2.first;
+        if (tlen != 0) {
+          tlen += (tlen > 0) ? 1 : -1;
+        }
 
-        printf("%s\t%d\t%s\t%d\t255\t%s\t=\t%d\t%d\t%s\t%s\tNH:i:%d\n", seq2->name.s, f2 & 0xFFFF, index.target_names_[tr].c_str(), posread, cig, posmate, (x1.first-x2.first), (f2 & 0x10) ? &buf1[0] : seq2->seq.s,  (f2 & 0x10) ? &buf2[0] : seq2->qual.s, nmap);
+        printf("%s\t%d\t%s\t%d\t255\t%s\t=\t%d\t%d\t%s\t%s\tNH:i:%d\n", seq2->name.s, f2 & 0xFFFF, index.target_names_[tr].c_str(), posread, cig, posmate, tlen, (f2 & 0x10) ? &buf1[0] : seq2->seq.s,  (f2 & 0x10) ? &buf2[0] : seq2->qual.s, nmap);
       }
 
 
@@ -471,7 +477,7 @@ void outputPseudoBam(const KmerIndex &index, int ec, const kseq_t *seq1, const s
           }
         }
 
-        int posread = (f1 & 0x10) ? (x1.first - seq1->seq.l) : x1.first;
+        int posread = (f1 & 0x10) ? (x1.first - seq1->seq.l+1) : x1.first;
         int dummy=1;
         getCIGARandSoftClip(cig, bool(f1 & 0x10), (f1 & 0x04) == 0, posread, dummy, seq1->seq.l, index.target_lens_[tr]);