Revert "Implementing #2045"

This reverts commit ee1f61b.

code/ARAX/ARAXQuery/ARAX_expander.py

@@ -697,8 +697,7 @@ def apply(self, response, input_parameters, mode: str = "ARAX"):
 
         # Map canonical curies back to the input curies in the QG (where applicable) #1622
         self._map_back_to_input_curies(message.knowledge_graph, query_graph, log)
-        eu.remove_semmeddb_edges_and_nodes_with_low_publications(message.knowledge_graph, response)
-        overarching_kg = eu.convert_standard_kg_to_qg_organized_kg(message.knowledge_graph)
+
         # Return the response and done
         kg = message.knowledge_graph
         log.info(f"After Expand, the KG has {len(kg.nodes)} nodes and {len(kg.edges)} edges "

code/ARAX/ARAXQuery/Expand/expand_utilities.py

@@ -555,53 +555,6 @@ def remove_edges_with_qedge_key(kg: KnowledgeGraph, qedge_key: str):
         if qedge_key in edge.qedge_keys:
             del kg.edges[edge_key]
 
-def remove_semmeddb_edges_and_nodes_with_low_publications(kg: KnowledgeGraph, log: ARAXResponse):
-        publication_threshold = 4
-        edge_keys = set(kg.edges)
-        edges_removed_counter = 0
-        removed_nodes = set()
-        connected_nodes = set()
-        try:
-            for edge_key in edge_keys:
-                edge = kg.edges[edge_key]
-                if not (edge.sources and any(retrieval_source.resource_id == 'infores:semmeddb' and
-                                                                    retrieval_source.resource_role == "primary_knowledge_source"
-                                                                    for retrieval_source in edge.sources)):
-                    connected_nodes.add(edge.subject)
-                    connected_nodes.add(edge.object)
-                    continue
-                if not edge.attributes:
-                    removed_nodes.add(edge.subject)
-                    removed_nodes.add(edge.object)
-                    del kg.edges[edge_key]
-                    edges_removed_counter += 1
-                    continue
-                n_publications = 0
-                for attribute in edge.attributes:
-                    if attribute.attribute_type_id == 'biolink:publications':
-                        if isinstance(attribute.value, list):
-                            n_publications = len(attribute.value)
-                if n_publications < publication_threshold:
-                    removed_nodes.add(edge.subject)
-                    removed_nodes.add(edge.object)
-                    del kg.edges[edge_key]
-                    edges_removed_counter += 1
-                else:
-                    connected_nodes.add(edge.subject)
-                    connected_nodes.add(edge.object)
-            orphaned_nodes = removed_nodes - connected_nodes
-            for node_key in orphaned_nodes:
-                del kg.nodes[node_key]
-        except:
-            tb = traceback.format_exc()
-            error_type, error, _ = sys.exc_info()
-            log.error(tb, error_code=error_type.__name__)
-            log.error(f"Something went wrong removing semmeddb edges from the knowledge graph")
-        else:
-            log.info(f"{edges_removed_counter} Semmeddb Edges with low publications successfully removed")
-
-
-
 
 def is_expand_created_subclass_qedge_key(qedge_key: str, qg: QueryGraph) -> bool:
     """

code/ARAX/test/test_ARAX_expand.py

@@ -793,8 +793,7 @@ def test_qualified_regulates_query():
     query = {
         "nodes": {
             "n0": {
-                "ids": ["NCBIGene:7157"]
-                # "ids": ["NCBIGene:375"]
+                "ids": ["NCBIGene:375"]
             },
             "n1": {
                 "categories": ["biolink:Gene"]
@@ -830,6 +829,7 @@ def test_qualified_regulates_query():
     }
     nodes_by_qg_id, edges_by_qg_id = _run_query_and_do_standard_testing(json_query=query)
 
+
 def test_1516_single_quotes_in_ids():
     actions = [
         "add_qnode(key=n0,ids=UniProtKB:P00491)",

code/ARAX/test/test_ARAX_filter_results.py

@@ -54,12 +54,12 @@ def test_n_results():
             "expand(edge_key=e00, kp=infores:rtx-kg2)",
             "overlay(action=add_node_pmids, max_num=15)",
             "resultify(ignore_edge_direction=true)",
-            "filter_results(action=sort_by_node_attribute, node_attribute=pubmed_ids, direction=a, max_results=5)",
+            "filter_results(action=sort_by_node_attribute, node_attribute=pubmed_ids, direction=a, max_results=20)",
             "return(message=true, store=false)"
         ]}}
     [response, message] = _do_arax_query(query)
     assert response.status == 'OK'
-    assert message.n_results == len(message.results) == 5
+    assert message.n_results == len(message.results) == 20
 
 def test_no_results():
     query = {"operations": {"actions": [
@@ -128,13 +128,13 @@ def test_warning():
             "expand(edge_key=e00, kp=infores:rtx-kg2)",
             "overlay(action=add_node_pmids, max_num=15)",
             "resultify(ignore_edge_direction=true)",
-            "filter_results(action=sort_by_node_attribute, node_attribute=pubmed_ids, direction=a, max_results=5)",
-            "filter_results(action=sort_by_node_attribute, node_attribute=asdfghjkl, direction=a, max_results=5)",
+            "filter_results(action=sort_by_node_attribute, node_attribute=pubmed_ids, direction=a, max_results=20)",
+            "filter_results(action=sort_by_node_attribute, node_attribute=asdfghjkl, direction=a, max_results=20)",
             "return(message=true, store=false)"
         ]}}
     [response, message] = _do_arax_query(query)
     assert response.status == 'OK'
-    assert len(message.results) == 5
+    assert len(message.results) == 20
 
 @pytest.mark.slow
 def test_sort_by_edge_attribute():
@@ -166,12 +166,12 @@ def test_sort_by_node_attribute():
             "expand(edge_key=e00, kp=infores:rtx-kg2)",
             "overlay(action=add_node_pmids, max_num=15)",
             "resultify(ignore_edge_direction=true)",
-            "filter_results(action=sort_by_node_attribute, node_attribute=pubmed_ids, direction=a, max_results=5, qnode_keys=[n01])",
+            "filter_results(action=sort_by_node_attribute, node_attribute=pubmed_ids, direction=a, max_results=20, qnode_keys=[n01])",
             "return(message=true, store=false)"
         ]}}
     [response, message] = _do_arax_query(query)
     assert response.status == 'OK'
-    assert len(message.results) == 5
+    assert len(message.results) == 20
     # add something to test if the results are assending and the correct numbers
 
 def test_sort_by_score():
@@ -182,12 +182,12 @@ def test_sort_by_score():
             "add_qedge(subject=n00, object=n01, key=e00)",
             "expand(edge_key=e00, kp=infores:rtx-kg2)",
             "resultify(ignore_edge_direction=true)",
-            "filter_results(action=sort_by_score, direction=a, max_results=8)",
+            "filter_results(action=sort_by_score, direction=a, max_results=20)",
             "return(message=true, store=false)"
         ]}}
     [response, message] = _do_arax_query(query)
     assert response.status == 'OK'
-    assert len(message.results) == 8
+    assert len(message.results) == 20
     result_scores = [x.analyses[0].score for x in message.results]
     assert result_scores == sorted(result_scores)
     assert max(result_scores) < 1