Merge pull request #4446 from xchang1/distanceless-distance-index

Speed up building the distance index without distances

src/snarl_distance_index.cpp

@@ -436,6 +436,7 @@ SnarlDistanceIndex::TemporaryDistanceIndex make_temporary_distance_index(
         temp_snarl_record.node_count = temp_snarl_record.children.size();
     }
 
+
     /*Now go through the decomposition again to fill in the distances
      * This traverses all chains in reverse order that we found them in, so bottom up
      * Each chain and snarl already knows its parents and children, except for single nodes
@@ -1071,6 +1072,9 @@ void populate_snarl_index(
         temp_index.use_oversized_snarls = true;
     }
 
+    if (size_limit == 0) {
+        all_children.clear();
+    }
     //Add the start and end nodes to the list of children so that we include them in the traversal 
     if (!temp_snarl_record.is_root_snarl) {
         all_children.emplace_back(SnarlDistanceIndex::TEMP_NODE, temp_snarl_record.start_node_id);
@@ -1432,9 +1436,14 @@ void populate_snarl_index(
         }
     }
 
+    //If we aren't keeping track of distances, then we didn't actually go through the snarl so we don't know if the snarl was simple or not
+    if (size_limit == 0) {
+        temp_snarl_record.is_simple = false;
+    }
+
     //If this is a simple snarl (one with only single nodes that connect to the start and end nodes), then
     // we want to remember if the child nodes are reversed 
-    if (temp_snarl_record.is_simple) {
+    if (size_limit != 0 && temp_snarl_record.is_simple) {
         for (size_t i = 0 ; i < temp_snarl_record.node_count ; i++) {
             //Get the index of the child
             const pair<SnarlDistanceIndex::temp_record_t, size_t>& child_index = temp_snarl_record.children[i];

src/unittest/snarl_distance_index.cpp

@@ -150,7 +150,7 @@ namespace vg {
                 REQUIRE(std::get<2>(traceback.second.back()) == -5);
             }
         }
-        TEST_CASE( "Nested chain with loop", "[snarl_distance][bug]" ) {
+        TEST_CASE( "Nested chain with loop", "[snarl_distance]" ) {
 
             VG graph;
 
@@ -263,6 +263,11 @@ namespace vg {
             SECTION("Distanceless index") {
                 SnarlDistanceIndex distance_index;
                 fill_in_distance_index(&distance_index, &graph, &snarl_finder, 0);
+
+                for (auto& id : {n1->id(), n2->id(), n3->id(), n4->id(), n5->id(), n6->id(), n7->id(), n8->id(), n9->id(), n10->id(), n11->id(), n12->id(), n13->id()}) { 
+                    net_handle_t n = distance_index.get_node_net_handle(id);
+                    distance_index.get_parent(n);
+                }
             }
         }
         TEST_CASE( "Snarl decomposition can deal with multiple connected components",
@@ -3216,6 +3221,15 @@ namespace vg {
                     REQUIRE(distance_index.minimum_distance(
                              n5->id(), false, 0, n5->id(), true, 0) == std::numeric_limits<size_t>::max());
                 }
+                SECTION("Distanceless index") {
+                    SnarlDistanceIndex distance_index;
+                    fill_in_distance_index(&distance_index, &graph, &snarl_finder, 0);
+
+                    for (auto& id : {n1->id(), n2->id(), n3->id(), n4->id(), n5->id(), n6->id(), n7->id(), n8->id()}) { 
+                        net_handle_t n = distance_index.get_node_net_handle(id);
+                        distance_index.get_parent(n);
+                    }
+                }
 
             }