Merge pull request #4454 from vgteam/distanceless-index

Distanceless index

src/snarl_distance_index.cpp

@@ -1072,9 +1072,6 @@ 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);
@@ -1141,8 +1138,9 @@ void populate_snarl_index(
           //}
 
         if ( (temp_snarl_record.node_count > size_limit || size_limit == 0) && (temp_snarl_record.is_root_snarl || (!start_is_tip &&
-             !start_rank == 0 && ! start_rank == 1))) {
+             start_rank != 0 && start_rank != 1))) {
             //If we don't care about internal distances, and we also are not at a boundary or tip
+            //TODO: Why do we care about tips specifically?
             continue;
         }
 
@@ -1213,7 +1211,23 @@ void populate_snarl_index(
 #endif
                 graph->follow_edges(current_end_handle, false, [&](const handle_t next_handle) {
                     if (graph->get_id(current_end_handle) == graph->get_id(next_handle)){
-                        //If there are any loops then this isn't a simple snarl
+                        //If this loops onto the same node side then this isn't a simple snarl
+                        temp_snarl_record.is_simple = false;
+                    } else if ((current_index.first == SnarlDistanceIndex::TEMP_NODE ? current_index.second 
+                                                                                     : (current_rev ? temp_index.temp_chain_records[current_index.second].end_node_id
+                                                                                                    : temp_index.temp_chain_records[current_index.second].start_node_id))
+                                    == graph->get_id(next_handle)){
+                        //If this loops to the other end of the chain then this isn't a simple snarl
+                        temp_snarl_record.is_simple = false;
+                    } else if (!temp_snarl_record.is_root_snarl && start_rank == 0 && 
+                               current_index != start_index && graph->get_id(next_handle) != temp_snarl_record.end_node_id) {
+                        //If the starting point of this traversal was the start of the snarl, the current starting point is not the start node,
+                        //and we found another child, then this is not a simple snarl
+                        temp_snarl_record.is_simple = false;
+                    } else if (!temp_snarl_record.is_root_snarl && start_rank == 1 && 
+                               current_index != start_index && graph->get_id(next_handle) != temp_snarl_record.start_node_id) {
+                        //If the starting point of this traversal was the end of the snarl, the current starting point is not the end node,
+                        //and we found another child, then this is not a simple snarl
                         temp_snarl_record.is_simple = false;
                     }
 
@@ -1328,7 +1342,7 @@ void populate_snarl_index(
                             }
                         } else if (!next_is_boundary && !temp_snarl_record.distances.count(make_pair(start, next))) {
                             //Otherwise the snarl stores it in its distance
-                            //If the distance isn't from an internal node to a bound and we haven't stored the distance yets
+                            //If the distance isn't from an internal node to a bound and we haven't stored the distance yet
 
                             temp_snarl_record.distances[make_pair(start, next)] = current_distance;
                             added_new_distance = true;
@@ -1436,14 +1450,10 @@ 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 (size_limit != 0 && temp_snarl_record.is_simple) {
+    if (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

@@ -266,7 +266,14 @@ namespace vg {
 
                 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);
+                    net_handle_t parent = distance_index.get_parent(n);
+                    try {
+                        distance_index.minimum_length(n);
+                        distance_index.minimum_length(parent);
+                        REQUIRE(false);
+                    } catch (const std::runtime_error& err) {
+                        REQUIRE(true);
+                    }
                 }
             }
         }