Making output deterministic

src/cgi/include/cgid_types.hpp

@@ -32,8 +32,9 @@ namespace cgi
   {
     bool operator() (const MappingResult_CGI &x, const MappingResult_CGI &y)
     {
-      return std::tie(x.genomeId, x.querySeqId, x.nucIdentity) 
-        < std::tie(y.genomeId, y.querySeqId, y.nucIdentity);
+      return std::tie(x.genomeId, x.querySeqId, x.nucIdentity, x.mapRefPosBin) 
+        < std::tie(y.genomeId, y.querySeqId, y.nucIdentity, y.mapRefPosBin);
+      //Added ref. bin also to make sort output deterministic [issue #46]
     }
   } cmp_query_bucket;
 

src/map/include/winSketch.hpp

@@ -49,7 +49,7 @@ namespace skch
       const skch::Parameters &param;
 
       //Ignore top % most frequent minimizers while lookups
-      const float percentageThreshold = 0.001;
+      const float percentageThreshold = 0.0;
 
       //Minimizers that occur this or more times will be ignored (computed based on percentageThreshold)
       int freqThreshold = std::numeric_limits<int>::max();
@@ -123,6 +123,9 @@ namespace skch
         //sequence counter while parsing file
         seqno_t seqCounter = 0;
 
+        if ( omp_get_thread_num() == 0)
+          std::cerr << "INFO [thread 0], skch::Sketch::build, window size for minimizer sampling  = " << param.windowSize << std::endl;
+
         for(const auto &fileName : param.refSequences)
         {
 
@@ -238,7 +241,7 @@ namespace skch
         else
         {
           if ( omp_get_thread_num() == 0)
-            std::cerr << "INFO [thread 0], skch::Sketch::computeFreqHist, With threshold " << this->percentageThreshold << "\%, consider all minimizers during lookup." << std::endl;
+            std::cerr << "INFO [thread 0], skch::Sketch::computeFreqHist, consider all minimizers during lookup." << std::endl;
         }
 
       }