From 6bfe0b4bdcc21dea7b4f79a3698339c3e4000a17 Mon Sep 17 00:00:00 2001
From: Kevin Vizhalil <kvnthomas98@gmail.com>
Date: Mon, 7 Oct 2024 09:41:24 -0400
Subject: [PATCH] split patterns and synonyms #2388

---
 code/ARAX/ARAXQuery/Filter_KG/remove_nodes.py | 21 +++++++++++--------
 .../KnowledgeSources/general_concepts.json    |  8 +++----
 2 files changed, 16 insertions(+), 13 deletions(-)

diff --git a/code/ARAX/ARAXQuery/Filter_KG/remove_nodes.py b/code/ARAX/ARAXQuery/Filter_KG/remove_nodes.py
index fbe66f886..53569fe7b 100644
--- a/code/ARAX/ARAXQuery/Filter_KG/remove_nodes.py
+++ b/code/ARAX/ARAXQuery/Filter_KG/remove_nodes.py
@@ -157,23 +157,25 @@ def remove_orphaned_nodes(self):
         return self.response
     
     def _is_general_concept(self, node):
-        curies = []
-        synonyms = []
+        curies = set()
+        synonyms = set()
         if not node['attributes']:
             return False
         for attribute in node['attributes']:
             if attribute['attribute_type_id'] == 'biolink:xref':
-                curies += attribute.get('value',[])
+                curies.update(map(str.lower, attribute.get('value', [])))
             if attribute['attribute_type_id'] == 'biolink:synonym':
-                synonyms += attribute.get('value',[])
+                synonyms.update(map(str.lower, attribute.get('value', [])))
         if node['name']:
-                synonyms.append(node['name'].lower())
-        if self.block_list_curies.intersection([curie.lower() for curie in curies if curie]):
+                synonyms.add(node['name'].lower())
+        if self.block_list_curies.intersection(curies) or self.block_list_synonyms.intersection(synonyms):
             return True
+        
         for synonym in synonyms:
-            for block_list_synonym in self.block_list_synonyms:
-                if isinstance(synonym,str) and isinstance(block_list_synonym,str) and re.match(block_list_synonym, synonym,re.IGNORECASE):
-                    return True
+            if not isinstance(synonym,str):
+                continue 
+            if any(p.match(synonym) for p in self.block_list_patterns):
+                return True
         return False
     
     def remove_general_concept_nodes(self):
@@ -200,6 +202,7 @@ def remove_general_concept_nodes(self):
             self.block_list_synonyms = set(block_list_dict["synonyms"])
             self.block_list_curies = set(block_list_dict["curies"])
             node_to_remove = set()
+            self.block_list_patterns = [re.compile(pattern,re.IGNORECASE) for pattern in block_list_dict["patterns"]]
             # iterate over edges find edges connected to the nodes
             for key, edge in self.message.knowledge_graph.edges.items():
                 if set({edge.subject, edge.object}).intersection(node_to_remove):
diff --git a/code/ARAX/KnowledgeSources/general_concepts.json b/code/ARAX/KnowledgeSources/general_concepts.json
index 4e30b2e89..79709da75 100644
--- a/code/ARAX/KnowledgeSources/general_concepts.json
+++ b/code/ARAX/KnowledgeSources/general_concepts.json
@@ -347,7 +347,6 @@
         "umls:c0003209",
         "umls:c4045974",
         "umls:c0005522"
-
     ],
     "synonyms": [
         "used in nicotine dependence",
@@ -684,7 +683,6 @@
         "secondary",
         "uterotonics",
         "radiotherapy",
-        "pharmacolog.*",
         "medicament",
         "Anesthetics",
         "vaccines",
@@ -708,6 +706,8 @@
         "Introns",
         "antioxidant",
         "Oils"
-
+    ],
+    "patterns": [
+        "pharmacolog.*"
     ]
-}
+}
\ No newline at end of file