fix chunking

src/convert.h

@@ -528,7 +528,7 @@ namespace lorax
       }
 
       // Clean-up
-      std::cerr << '[' << boost::posix_time::to_simple_string(boost::posix_time::second_clock::local_time()) << "] Processed " << aln.size() << " alignments" << std::endl;
+      if (c.chunk) std::cerr << '[' << boost::posix_time::to_simple_string(boost::posix_time::second_clock::local_time()) << "] Processed " << processed.size() << " reads" << std::endl;
       firstPass = false;
       aln.clear();
     } while (!gafdone);
@@ -551,7 +551,7 @@ namespace lorax
     boost::program_options::options_description generic("Generic options");
     generic.add_options()
       ("help,?", "show help message")
-      ("chunk,c", boost::program_options::value<uint32_t>(&c.chunk)->default_value(0), "chunk size [0: all at once]")
+      ("chunk,c", boost::program_options::value<uint32_t>(&c.chunk)->default_value(500000), "chunk size [0: all at once]")
       ("graph,g", boost::program_options::value<boost::filesystem::path>(&c.gfafile), "GFA pan-genome graph")
       ("reference,r", boost::program_options::value<boost::filesystem::path>(&c.genome), "FASTA reference")
       ("align,a", boost::program_options::value<boost::filesystem::path>(&c.readsfile), "BAM/CRAM file")

src/gaf.h

@@ -188,15 +188,15 @@ namespace lorax
 	      }
 	      newReads.insert(ar.seed);
 	    }
-	  }
+	  } else continue; // Already processed
 	}
 	aln.push_back(ar);
 	//std::cerr << ar.seed << ',' << ar.qlen << ',' << ar.qstart << ',' << ar.qend << ',' << ar.strand << ',' << ar.plen << ',' << ar.pstart << ',' << ar.pend << ',' << ar.matches << ',' << ar.alignlen << ',' << ar.mapq << std::endl;
       }
       else parseAR = false;
     }
     // Keep track of processed reads
-    if (!maxread) seedSet.insert(newReads.begin(), newReads.end());
+    if (maxread) seedSet.insert(newReads.begin(), newReads.end());
 
     // Close file
     dataIn.pop();

src/lorax.cpp

@@ -45,7 +45,7 @@ displayUsage() {
   std::cout << "    convert       convert pan-genome graph alignment to BAM" << std::endl;
   std::cout << "    gfa2dot       convert pan-genome graph to dot (graphviz) format" << std::endl;
   std::cout << "    ncov          node coverage" << std::endl;
-  std::cout << "    geno          edge-genotyping using pan-genome alignments" << std::endl;
+  //std::cout << "    geno          edge-genotyping using pan-genome alignments" << std::endl;
   std::cout << std::endl;
   std::cout << std::endl;
 }