Disable symmetrizeOverlaps() for layoutReads.

src/bogart/AS_BAT_OverlapCache.C

@@ -41,7 +41,8 @@ OverlapCache::OverlapCache(const char *ovlStorePath,
                            double maxErate,
                            uint32 minOverlap,
                            uint64 memlimit,
-                           uint64 genomeSize) {
+                           uint64 genomeSize,
+                           bool symmetrize) {
 
   _prefix = prefix;
 
@@ -159,7 +160,8 @@ OverlapCache::OverlapCache(const char *ovlStorePath,
   delete [] _ovsTmp;    _ovsTmp   = NULL;   //  it loaded updated erates into memory), so release
   delete     ovlStore;   ovlStore = NULL;   //  these before symmetrizing overlaps.
 
-  symmetrizeOverlaps();
+  if (symmetrize == true)
+    symmetrizeOverlaps();
 
   delete [] _minSco;    _minSco   = NULL;
 }
@@ -288,21 +290,6 @@ OverlapCache::computeOverlapLimit(ovStore *ovlStore, uint64 genomeSize) {
     _maxPer += increase;
   }
 
-  //  We used to (pre 6 Jul 2017) do the symmetry check only if we didn't load all overlaps.
-  //  However, symmetry can also break if we use an error rate cutoff because - for reasons not
-  //  explored - the error rate on symmetric overlaps differs.  So, just enable this always.
-  //
-  //  On a moderate coverage human nanopore assembly, it does:
-  //
-  //    OverlapCache()-- Symmetrizing overlaps -- finding missing twins.
-  //    OverlapCache()--                       -- found 8609 missing twins in 51413413 overlaps, 8002 are strong.
-  //    OverlapCache()-- Symmetrizing overlaps -- dropping weak non-twin overlaps.
-  //    OverlapCache()--                       -- dropped 454 overlaps.
-  //    OverlapCache()-- Symmetrizing overlaps -- adding 8155 missing twin overlaps.
-
-  _checkSymmetry = (numAbove > 0) ? true : false;
-  _checkSymmetry = true;
-
   if (_maxPer < _minPer)
     writeStatus("OverlapCache()-- Not enough memory to load the minimum number of overlaps; increase -M.\n"), exit(1);
 
@@ -650,9 +637,6 @@ OverlapCache::symmetrizeOverlaps(void) {
   uint32  numThreads = omp_get_max_threads();
   uint32  blockSize  = (fiLimit < 1000 * numThreads) ? numThreads : fiLimit / 999;
 
-  if (_checkSymmetry == false)
-    return;
-
   uint32  *nNonSymPerRead = new uint32 [fiLimit];
   uint32  *nFiltPerRead   = new uint32 [fiLimit];
   uint32  *nMissPerRead   = new uint32 [fiLimit];

src/bogart/AS_BAT_OverlapCache.H

@@ -244,7 +244,8 @@ public:
                double maxErate,
                uint32 minOverlap,
                uint64 maxMemory,
-               uint64 genomeSize);
+               uint64 genomeSize,
+               bool symmetrize=true);
   ~OverlapCache();
 
   bool         compareOverlaps(const BAToverlap &a, const BAToverlap &b) const; // we can almost do templated but the fields are functions in one and just members in the other
@@ -293,8 +294,6 @@ private:
 
   uint64                 *_minSco;     //  The minimum score accepted for each read
 
-  bool                    _checkSymmetry;
-
   uint32                  _ovsMax;     //  For loading overlaps
   ovOverlap              *_ovs;        //
   uint64                 *_ovsSco;     //  For scoring overlaps during the load

src/bogart/layoutReads.C

@@ -273,6 +273,8 @@ main (int argc, char **argv) {
   uint64       genomeSize         = 0;
   bool         doContainPlacement = true;
 
+  bool         doSymmetrize       = false;
+
   argc = AS_configure(argc, argv);
 
   uint32 seed = time(NULL);
@@ -315,6 +317,10 @@ main (int argc, char **argv) {
       doContainPlacement = false;
     }
 
+    else if (strcmp(argv[arg], "-symmetrize") == 0) {
+      doSymmetrize = true;
+    }
+
     else {
       char *s = new char [1024];
       snprintf(s, 1024, "Unknown option '%s'.\n", argv[arg]);
@@ -344,6 +350,8 @@ main (int argc, char **argv) {
     fprintf(stderr, "  -eM erate          Max error rate of overlaps to load.\n");
     fprintf(stderr, "  -eg erate          Max error rate of overlaps to use for placing contained reads.\n");
     fprintf(stderr, "  -nocontains        Do not place contained reads.\n");
+    fprintf(stderr, "\n");
+    fprintf(stderr, "  -symmetrize        Check and fix symmetry of overlaps (slow).\n");
 
     for (uint32 ii=0; ii<err.size(); ii++)
       if (err[ii])
@@ -361,7 +369,7 @@ main (int argc, char **argv) {
   setLogFile(prefix, "loadInformation");
 
   RI = new ReadInfo(seqStorePath, prefix, minReadLen, maxReadLen);
-  OC = new OverlapCache(ovlStorePath, prefix, std::max(erateMax, erateGraph), minOverlapLen, ovlCacheMemory, genomeSize);
+  OC = new OverlapCache(ovlStorePath, prefix, std::max(erateMax, erateGraph), minOverlapLen, ovlCacheMemory, genomeSize, doSymmetrize);
 
   //