adding the data

backend/app/api/endpoints/search_endpoint.py

@@ -35,7 +35,7 @@
 @version(1, 0)
 async def get_search_results(
     query: str = "",
-    routecategory: RouteCategory =RouteCategory.CT# RouteCategory.NS
+    routecategory: RouteCategory =RouteCategory.DRUG# RouteCategory.NS
 ) -> JSONResponse:
     if trace_transaction := sentry_sdk.Hub.current.scope.transaction:
         trace_transaction.set_tag("title", 'api_get_search_results')

backend/app/chebml_eval_data.csv

@@ -0,0 +1,11 @@
+Study Title,Question,Answer,Link,Source
+Aspirin,What is the alogP value for Aspirin?,The alogP value for Aspirin is 1.33.,CHEMBL6328,chembl
+Aspirin,Has Aspirin received a black box warning from the FDA?,"No, Aspirin has not received a black box warning from the FDA.",CHEMBL267864,chembl
+Aspirin,How many aromatic rings are present in the Aspirin molecule?,The Aspirin molecule contains 3 aromatic rings.,CHEMBL6329,chembl
+Aspirin,How many aromatic rings are present in the structure of Aspirin?,Aspirin has 3 aromatic rings.,CHEMBL267864,chembl
+Aspirin,How many aromatic rings are present in the Aspirin molecule?,The Aspirin molecule has 3 aromatic rings.,CHEMBL6362,chembl
+Paracetamol (Acetaminophen),What is the molecular type of Paracetamol (Acetaminophen)?,The molecular type of Paracetamol (Acetaminophen) is a 'Small molecule'.,CHEMBL6362,chembl
+Paracetamol (Acetaminophen),What is the chirality of Paracetamol (Acetaminophen)?,The chirality of Paracetamol (Acetaminophen) is -1.,CHEMBL265667,chembl
+Paracetamol (Acetaminophen),What is the alogp value for Paracetamol (Acetaminophen)?,The alogp value for Paracetamol (Acetaminophen) is 1.33.,CHEMBL6328,chembl
+Paracetamol (Acetaminophen),How many aromatic rings are present in the structure of Paracetamol (Acetaminophen)?,Paracetamol (Acetaminophen) has three aromatic rings in its structure.,CHEMBL265667,chembl
+Paracetamol (Acetaminophen),What is the value of alogP for Paracetamol (Acetaminophen)?,The value of alogP for Paracetamol (Acetaminophen) is 2.11.,CHEMBL6329,chembl

backend/app/make_it_rain.py

@@ -0,0 +1,41 @@
+# %%
+import requests
+from tqdm import tqdm
+import pandas as pd 
+import json 
+#%%
+df = pd.read_csv("chebml_eval_data.csv")
+
+#%%
+df_filter = df[df['Source'] == "Clicnical Trials"].sample(n = 10)
+chembl_data_debug = pd.DataFrame()
+for i in tqdm(df_filter.iterrows()): 
+    url = f"http://127.0.0.1:8000/search?query={i[1]['Question']}"
+    data_store = {
+        "question": i[1]['Question'],
+        "answer": i[1]['Answer'], 
+        "study": i[1]['Study Title'],
+    }
+    payload = {}
+    headers = {
+    'Authorization': 'Bearer eyJhbGciOiJIUzI1NiIsInR5cCI6IkpXVCJ9.eyJzdWIiOiJjdXJpZW8iLCJqdGkiOiI2ZmI2MTEyNS1jZGU1LTQ2MDAtYWE2MS1jMjBiYzEwNmRhNDMiLCJ0eXBlIjoiYWNjZXNzIiwiZnJlc2giOmZhbHNlLCJpYXQiOjE3MTA3NjUzMDYsImV4cCI6MTcxMDc2NjIwNi4yNzg4Mzd9.nmuzrzmr81ulI8TDauwx19QvLHFi8nXJeUgrEVzfhXs'
+    }
+    tries = 0
+    while tries < 1: 
+        try: 
+            response = requests.request("GET", url, headers=headers, data=payload)
+            json_str = response.text.replace("'", '"').strip('"')
+            data = json.loads(json_str)
+            if len(data['results']) > 0:
+                data_store['refined_results'] = data['result']
+                break
+        except: 
+            tries += 1 
+            continue
+    if data_store.get("refined_results", None) is not None:
+        chembl_data_debug = pd.concat([chembl_data_debug, pd.DataFrame([data_store])], ignore_index=  True )
+    else: 
+        data_store['refined_results'] = ""
+        chembl_data_debug = pd.concat([chembl_data_debug, pd.DataFrame([data_store])], ignore_index= True)
+
+chembl_data_debug.to_csv("chembl_data_debug_source.csv", index = False)

backend/app/rag/retrieval/drug_chembl/drug_chembl_graph_query_engine.py

@@ -72,8 +72,8 @@ def __init__(self, config):
             "Cypher Response: {context_str}\n"
             "Response: "
         )
-
         self.qp = self.build_query_pipeline()
+        self.debug_chembl = {}
 
     def execute_graph_query(self, queries):
         logger.info(
@@ -110,7 +110,8 @@ def execute_graph_query(self, queries):
 
             result_dict = self.graph_storage.execute_query(query)
             results.append(result_dict)
-
+
+        self.debug_chembl['cypher_query'] = str(query_list)
         logger.info(
             f"execute_graph_query results: {results}"
         )
@@ -156,9 +157,15 @@ def get_table_context_str(self, table_schema_objs: List[dict[str, str]]) -> str:
                 table_context += f"{key}: {value}\n"
 
             context_strs.append(table_context)
+        self.debug_chembl['table_context_str'] = "\n\n".join(context_strs)
 
         return "\n\n".join(context_strs)
 
+
+    def store_debug(self): 
+        with open("debug_chembl.txt", "w") as f:
+            f.write(str(self.debug_chembl) + ",\n")
+
     def get_response_synthesis_prompt(
         self, query_str, sql_query, context_str
     ) -> PromptTemplate:
@@ -181,6 +188,7 @@ def cypher_output_parser(self, response: list[dict[str, list]]) -> str:
 
             response_str += " ## ".join(record_in_list) + "\n"
 
+        self.debug_chembl['cypher_response'] = response_str
         logger.info(
             f"cypher_output_parser response_str: {response_str}"
         )
@@ -233,11 +241,13 @@ def build_query_pipeline(self):
     async def call_text2cypher(self, search_text:str) -> str:
         try:
             logger.info(f"call_text2cypher search_text: {search_text}")
-
+            self.debug_chembl['question'] = search_text
             response = self.qp.run(query=search_text)
 
-            logger.info(f"call_text2cypher response: {str(response)}")
+            self.debug_chembl['response'] = response
 
+            logger.info(f"call_text2cypher response: {str(response)}")
+            self.store_debug()
         except Exception as ex:
             logger.exception("call_text2cypher Exception -", exc_info = ex, stack_info=True)
 

backend/app/router/orchestrator.py

@@ -40,9 +40,9 @@ def __init__(self, config):
         self.router.load(ORCHESRATOR_ROUTER_PROMPT_PROGRAM)
 
         self.clinicalTrialSearch = ClinicalTrialText2SQLEngine(config)
-        # self.drugChemblSearch = DrugChEMBLText2CypherEngine(config)
-        # self.pubmedsearch = PubmedSearchQueryEngine(config)
-        # self.bravesearch = BraveSearchQueryEngine(config)
+        self.drugChemblSearch = DrugChEMBLText2CypherEngine(config)
+        self.pubmedsearch = PubmedSearchQueryEngine(config)
+        self.bravesearch = BraveSearchQueryEngine(config)
 
     async def query_and_get_answer(
         self,
@@ -87,31 +87,31 @@ async def query_and_get_answer(
                 logger.exception("Orchestrator.query_and_get_answer.sqlResponse Exception -", exc_info = e, stack_info=True)
                 pass
 
-        # elif router_id == 1 or routecategory == RouteCategory.DRUG:
-        #     # drug information call
-        #     logger.info(
-        #         "Orchestrator.query_and_get_answer.router_id drug_information_choice Entered."
-        #     )
-        #     try:
-        #         cypherResponse = await self.drugChemblSearch.call_text2cypher(
-        #             search_text=search_text
-        #         )
-        #         result = str(cypherResponse)
-        #         sources = result
-        #         logger.info(
-        #             f"Orchestrator.query_and_get_answer.cypherResponse cypherResponse: {result}"
-        #         )
-
-        #         return {
-        #             "result" : result,
-        #             "sources": sources
-        #         }
-        #     except Exception as e:
-        #         logger.exception(
-        #             "Orchestrator.query_and_get_answer.cypherResponse Exception -",
-        #             exc_info=e,
-        #             stack_info=True,
-        #         )
+        elif router_id == 1 or routecategory == RouteCategory.DRUG:
+            # drug information call
+            logger.info(
+                "Orchestrator.query_and_get_answer.router_id drug_information_choice Entered."
+            )
+            try:
+                cypherResponse = await self.drugChemblSearch.call_text2cypher(
+                    search_text=search_text
+                )
+                result = str(cypherResponse)
+                sources = result
+                logger.info(
+                    f"Orchestrator.query_and_get_answer.cypherResponse cypherResponse: {result}"
+                )
+
+                return {
+                    "result" : result,
+                    "sources": sources
+                }
+            except Exception as e:
+                logger.exception(
+                    "Orchestrator.query_and_get_answer.cypherResponse Exception -",
+                    exc_info=e,
+                    stack_info=True,
+                )
 
 
         # if routing fails, sql and cypher calls fail, routing to pubmed or brave