Merge pull request #136 from PolinaBevad/f_fix_string_out_of_bound_three_del_ins

Fix StringOutOfBoundException for cases of three indels/three deletions/two deletions

Author: pcingola

src/main/java/com/astrazeneca/vardict/VarDict.java

@@ -524,6 +524,9 @@ static int somdict(Region segs, Map<Integer, Vars> vars1, Map<Integer, Vars> var
         Collections.sort(pp);
 
         for (Integer p : pp) {
+            if (p < segs.start || p > segs.end) {
+                continue;
+            }
             Vars v1 = vars1.get(p);
             Vars v2 = vars2.get(p);
             if (v1 == null && v2 == null) { // both samples have no coverage

src/main/java/com/astrazeneca/vardict/modules/CigarUtils.java

@@ -194,7 +194,7 @@ public static Tuple.Tuple2<Integer, String> modifyCigar(int indel,
 
                 //number of bases after refoff/rdoff that match in reference and read
                 int rn = 0;
-                while (isHasAndEquals(querySeq.charAt(rdoff + rn), ref, refoff + rn)) {
+                while (rdoff + rn < querySeq.length() && isHasAndEquals(querySeq.charAt(rdoff + rn), ref, refoff + rn)) {
                     rn++;
                 }
                 RDOFF += rn;
@@ -247,7 +247,7 @@ public static Tuple.Tuple2<Integer, String> modifyCigar(int indel,
 
                 //number of bases after refoff/rdoff that match in reference and read
                 int rn = 0;
-                while (isHasAndEquals(querySeq.charAt(rdoff + rn), ref, refoff + rn)) {
+                while (rdoff + rn < querySeq.length() && isHasAndEquals(querySeq.charAt(rdoff + rn), ref, refoff + rn)) {
                     rn++;
                 }
                 RDOFF += rn;
@@ -301,7 +301,7 @@ public static Tuple.Tuple2<Integer, String> modifyCigar(int indel,
                 }
 
                 int rn = 0;
-                while (isHasAndEquals(querySeq.charAt(rdoff + rn), ref, refoff + rn)) {
+                while (rdoff + rn < querySeq.length() && isHasAndEquals(querySeq.charAt(rdoff + rn), ref, refoff + rn)) {
                     rn++;
                 }
                 RDOFF += rn;

src/main/java/com/astrazeneca/vardict/modules/StructuralVariantsProcessor.java

@@ -1102,7 +1102,7 @@ static void findINVdisc (Map<Integer, VariationMap<String, Variation>> hash,
                         ins = SequenceUtil.reverseComplement(joinRef(ref, bp, pe));
                     }
                     if (conf.y) {
-                        System.err.printf("  Found INV with discordant pairs only: cnt: %d Len: %d %d-%d<->%d-%d %s\n",
+                        System.err.printf("  Found INV with discordant pairs only 5': cnt: %d Len: %d %d-%d<->%d-%d %s\n",
                                 cnt, len, end, rstart, me, rms, ins);
                     }
                     final Variation vref = getVariation(hash, bp,"-" + len + "^" + ins);
@@ -1185,7 +1185,7 @@ static void findINVdisc (Map<Integer, VariationMap<String, Variation>> hash,
                         ins = SequenceUtil.reverseComplement(joinRef(ref, bp, pe));
                     }
                     if (conf.y) {
-                        System.err.printf("  Found INV with discordant pairs only: cnt: %d Len: %d %d-%d<->%d-%d %s\n",
+                        System.err.printf("  Found INV with discordant pairs only 3': cnt: %d Len: %d %d-%d<->%d-%d %s\n",
                                 cnt, len, me, rms, end, rstart, ins);
                     }
                     final Variation vref = getVariation(hash, bp,"-" + len + "^" + ins);