Merge pull request #101 from Bioinformatics/feature-MANTA-1146

MANTA-1146 Fix imprecise SV filtering when CIEND is a subset of CIPOS

CHANGELOG.md

@@ -8,6 +8,7 @@
 - Update minimum supported linux OS from Centos 5 to 6 (MANTA-1249)
 
 ### Fixed
+- Fix imprecise SV filtering when CIEND is a subset of CIPOS (MANTA-1146)
 - Ensure consistent BND pairs for translocations or RNA fusions are selected during vcf merging (MANTA-1243)
 - Fix manta to tolerate deserialization differences in boost above/below v1.58 (MANTA-1262)
   - Impact of issue was an (infrequent) assertion using boost v1.58+: `Assertion 'size() == rhs.size()' failed`

src/c++/lib/applications/GenerateSVCandidates/SVCandidateProcessor.cpp

@@ -19,6 +19,7 @@
 
 /// \file
 /// \author Chris Saunders
+/// \author Naoki Nariai
 ///
 
 #include "SVCandidateProcessor.hh"
@@ -182,7 +183,7 @@ writeSV(
             }
         }
 
-        // check min size for candidate output:
+        // check min size, or if it is IMPRECISE case where CIEND is a subset of CIPOS for candidate output:
         if (isSVBelowMinSize(sv,opt.scanOpt.minCandidateVariantSize))
         {
 #ifdef DEBUG_GSV

src/c++/lib/format/VcfWriterSV.cpp

@@ -20,6 +20,7 @@
 /// \file
 /// \author Chris Saunders
 /// \author Felix Schlesinger
+/// \author Naoki Nariai
 ///
 
 #include "format/VcfWriterSV.hh"
@@ -550,12 +551,6 @@ writeInvdel(
     pos += bpABkptAdjust;
     endPos += bpBBkptAdjust;
 
-    if (isImprecise)
-    {
-        // check against the rare IMPRECISE case arising when CIEND is a subset of CIPOS:
-        endPos=std::max(endPos,pos+1);
-    }
-
     if (pos<1) return;
 
     // get REF

src/c++/lib/manta/SVCandidateUtil.cpp

@@ -19,6 +19,7 @@
 
 /// \file
 /// \author Chris Saunders
+/// \author Naoki Nariai
 ///
 
 #include "manta/SVCandidateUtil.hh"
@@ -32,12 +33,36 @@ isSVBelowMinSize(
 {
     if (sv.bp1.interval.tid != sv.bp2.interval.tid) return false;
 
+    // check if the rare IMPRECISE case arising when CIEND is a subset of CIPOS,
+    // which is considered as small SV size
+    if (isInvalidBreakpointInterval(sv)) return true;
+
     const pos_t bpSize(std::abs(sv.bp1.interval.range.center_pos() - sv.bp2.interval.range.center_pos())-1);
     const pos_t insertSize(sv.insertSeq.size());
 
     return (std::max(bpSize,insertSize) < static_cast<pos_t>(minSize));
 }
 
+bool
+isInvalidBreakpointInterval(const SVCandidate& sv)
+{
+    using namespace EXTENDED_SV_TYPE;
+    const index_t svType(getExtendedSVType(sv));
+
+    // if SV is not translocation, and IMPRECISE, then check if the SV's CIEND is a subset of CIPOS
+    if (!isSVTransloc(svType) && sv.isImprecise())
+    {
+        const bool isBp1First(sv.bp1.interval.range.begin_pos()<=sv.bp2.interval.range.begin_pos());
+
+        const SVBreakend& bpA(isBp1First ? sv.bp1 : sv.bp2);
+        const SVBreakend& bpB(isBp1First ? sv.bp2 : sv.bp1);
+
+        return bpB.interval.range.center_pos() <= bpA.interval.range.center_pos();
+    }
+
+    return false;
+}
+
 bool
 isCis(const SVCandidate& sv)
 {

src/c++/lib/manta/SVCandidateUtil.hh

@@ -19,6 +19,7 @@
 
 /// \file
 /// \author Chris Saunders
+/// \author Naoki Nariai
 ///
 
 #pragma once
@@ -33,6 +34,12 @@ isSVBelowMinSize(
     const SVCandidate& sv,
     const unsigned minSize);
 
+/// returns true if the rare IMPRECISE case arising when CIEND is a subset of CIPOS,
+/// which is considered as small SV size
+bool
+isInvalidBreakpointInterval(
+        const SVCandidate& sv);
+
 /// returns true if the sv is in cis orientation, i.e same chromosome
 /// and a right open breakend to the left of a left open breakend
 ///