In alignment free assembly, enforce a minimum coverage for AnchorGrap…

…h edges.
paoloshasta · Feb 1, 2025 · 1c4ae3f · 1c4ae3f
1 parent ce5a43a
commit 1c4ae3f
Show file tree

Hide file tree

Showing 4 changed files with 16 additions and 9 deletions.
diff --git a/src/Mode3Assembler.cpp b/src/Mode3Assembler.cpp
@@ -386,6 +386,11 @@ shared_ptr<AssemblyGraph> Mode3Assembler::assembleConnectedComponent(
     ostream& orientedReadsCsv,
     bool debug)
 {
+    // EXPOSE WHEN CODE STABILIZES.
+    // Minimum anchor graph, edge coverage, only for alignment-free assembly.
+    const uint64_t minEdgeCoverageAlignmentFree = 3;
+
+
     performanceLog << timestamp << "Assembling connected component " <<
         componentId << " of " << componentOrientedReadIds.size() << endl;
     cout << timestamp << "Assembling connected component " <<
@@ -427,7 +432,10 @@ shared_ptr<AssemblyGraph> Mode3Assembler::assembleConnectedComponent(
 
     // Now we can create the AnchorGraph for this connected component.
     // The constructor generates the vertices and edges.
-    AnchorGraph anchorGraph(anchors(), anchorIds);
+    const uint64_t minEdgeCoverage = (
+        options.anchorCreationMethod == "FromMarkerKmers" ? minEdgeCoverageAlignmentFree :
+        0);
+    AnchorGraph anchorGraph(anchors(), anchorIds, minEdgeCoverage);
 
 
      cout << "The AnchorGraph for this connected component has " <<

diff --git a/src/mode3-AnchorGraph.cpp b/src/mode3-AnchorGraph.cpp
@@ -25,7 +25,7 @@ using namespace mode3;
 
 
 // Create the AnchorGraph and its vertices and edges given a vector of AnchorIds.
-AnchorGraph::AnchorGraph(const Anchors& anchors, span<const AnchorId> anchorIds) :
+AnchorGraph::AnchorGraph(const Anchors& anchors, span<const AnchorId> anchorIds, uint64_t minEdgeCoverage) :
     anchorIds(anchorIds)
 {
 
@@ -47,7 +47,7 @@ AnchorGraph::AnchorGraph(const Anchors& anchors, span<const AnchorId> anchorIds)
     vector<uint64_t> counts;
     for(uint64_t localAnchorId0=0; localAnchorId0<anchorIds.size(); localAnchorId0++) {
         const AnchorId anchorId0 = anchorIds[localAnchorId0];
-        anchors.findChildren(anchorId0, children, counts);
+        anchors.findChildren(anchorId0, children, counts, minEdgeCoverage);
         const uint64_t n = children.size();
         SHASTA_ASSERT(n == counts.size());
         for(uint64_t i=0; i<n; i++) {

diff --git a/src/mode3-AnchorGraph.hpp b/src/mode3-AnchorGraph.hpp
@@ -113,7 +113,7 @@ class shasta::mode3::AnchorGraph : public AnchorGraphBaseClass {
 
     // Create the AnchorGraph and its vertices and edges given the AnchorIds
     // for a conneced component of the global anchor graph.
-    AnchorGraph(const Anchors&, span<const AnchorId> anchorIds);
+    AnchorGraph(const Anchors&, span<const AnchorId> anchorIds, uint64_t minEdgeCoverage);
 
     // The AnchorIds of this AnchorGraph. Each of these corresponds to a vertex.
     // An index in this vector is called a local anchor id.

diff --git a/src/mode3-AssemblyGraphDetangle4.cpp b/src/mode3-AssemblyGraphDetangle4.cpp
@@ -9,7 +9,7 @@ using namespace mode3;
 
 
 void AssemblyGraph::run4(
-    uint64_t /* threadCount */,
+    uint64_t threadCount,
     bool /* assembleSequence */,
     bool debug)
 {
@@ -22,10 +22,6 @@ void AssemblyGraph::run4(
 
     write("A");
 
-    detangleVertices4();
-    write("B");
-
-#if 0
     // Bubble cleanup.
     compress();
     for(uint64_t iteration=0; ; iteration ++) {
@@ -44,6 +40,9 @@ void AssemblyGraph::run4(
         compress();
     }
 
+    write("B");
+
+#if 0
     // Clean up short superbubbles.
     cleanupSuperbubbles(false,
         options.assemblyGraphOptions.superbubbleLengthThreshold1,