Add DISGENET API integration as a new database tool

.env.example

@@ -4,6 +4,10 @@
 # Required: Anthropic API Key for Claude models
 ANTHROPIC_API_KEY=your_anthropic_api_key_here
 
+# Required if DISGENET API use:
+DISGENET_API_KEY=
+DISGENET_API_SERVER_HOST=https://api.disgenet.com
+
 # Optional: OpenAI API Key (if using OpenAI models)
 OPENAI_API_KEY=your_openai_api_key_here
 

biomni/agent/a1.py

@@ -2,13 +2,19 @@
 import inspect
 import os
 import re
+import sys
 from collections.abc import Generator
 from datetime import datetime
 from pathlib import Path
 from typing import Any, Literal, TypedDict
 
 import pandas as pd
 from dotenv import load_dotenv
+
+if os.path.exists(".env"):
+    load_dotenv(".env", override=False)
+    print("Loaded environment variables from .env")
+
 from langchain_core.messages import AIMessage, BaseMessage, HumanMessage, SystemMessage
 from langchain_core.prompts import ChatPromptTemplate
 from langgraph.checkpoint.memory import MemorySaver
@@ -43,10 +49,6 @@
     textify_api_dict,
 )
 
-if os.path.exists(".env"):
-    load_dotenv(".env", override=False)
-    print("Loaded environment variables from .env")
-
 
 class AgentState(TypedDict):
     messages: list[BaseMessage]
@@ -79,6 +81,10 @@ def __init__(
             commercial_mode: If True, excludes datasets that require commercial licenses or are non-commercial only
 
         """
+        # --- DISGENET API key preflight (warn early, before the agent tries to select DISGENET tools) ---
+        # Note: We never prompt in non-interactive contexts (e.g., Flask/gunicorn); we only warn.
+        self.disgenet_api_available = self._disgenet_api_key_check()
+
         # Use default_config values for unspecified parameters
         if path is None:
             path = default_config.path
@@ -222,6 +228,47 @@ def __init__(
         self.timeout_seconds = timeout_seconds  # 10 minutes default timeout
         self.configure()
 
+    _DISGENET_KEY_WARNED: bool = False
+    _DISGENET_KEY_PROMPTED: bool = False
+
+    def _disgenet_api_key_check(self) -> bool:
+        """One-time DISGENET_API_KEY check.
+
+        - If missing, prints a warning early during agent init.
+        - If stdin is a TTY, prompts once per process for a key (press Enter to skip).
+        """
+        existing = os.getenv("DISGENET_API_KEY")
+        if existing:
+            return True
+
+        if not A1._DISGENET_KEY_WARNED:
+            print(
+                "⚠️  DISGENET_API_KEY is not set. Please provide the DISGENET API KEY in the environment in order to get full access to DISGENET tools."
+            )
+            A1._DISGENET_KEY_WARNED = True
+
+        # Never block in non-interactive contexts (e.g., Flask/gunicorn, CI).
+        if not sys.stdin.isatty():
+            return False
+
+        # Avoid repeatedly prompting if the agent is re-initialized in a loop.
+        if A1._DISGENET_KEY_PROMPTED:
+            return False
+        A1._DISGENET_KEY_PROMPTED = True
+
+        try:
+            from getpass import getpass
+
+            entered = getpass("Enter DISGENET_API_KEY (leave blank to skip): ").strip()
+            if not entered:
+                return False
+            os.environ["DISGENET_API_KEY"] = entered
+            print("✅ DISGENET_API_KEY set for this process.")
+            return True
+        except Exception:
+            # If prompting fails for any reason, just keep going with the warning-only behavior.
+            return False
+
     def add_tool(self, api):
         """Add a new tool to the agent's tool registry and make it available for retrieval.
 

biomni/config.py

@@ -29,13 +29,13 @@ class BiomniConfig:
     """
 
     # Data and execution settings
-    path: str = "./data"
-    timeout_seconds: int = 600
+    path: str = os.getenv("BIOMNI_PATH", "./data")
 
-    # LLM settings (API keys still from environment)
-    llm: str = "claude-sonnet-4-5"
-    temperature: float = 0.7
+    # LLM settings (API keys still from environment): claude-3-5-sonnet-20241022 || gpt-4o-mini-2024-07-18
+    llm: str = os.getenv("BIOMNI_LLM", "gpt-5-mini-2025-08-07")
+    temperature: float = float(os.getenv("BIOMNI_TEMPERATURE", "1"))  # 0.7
 
+    timeout_seconds: int = 600
     # Tool settings
     use_tool_retriever: bool = True
 

biomni/env_desc.py

@@ -19,7 +19,6 @@
     "ddinter_hormonal.csv": "Drug-drug interactions for systemic hormonal preparations from DDInter 2.0 database.",
     "ddinter_respiratory.csv": "Drug-drug interactions for respiratory system drugs from DDInter 2.0 database.",
     "ddinter_various.csv": "Drug-drug interactions for various drugs from DDInter 2.0 database.",
-    "DisGeNET.parquet": "Gene-disease associations from multiple sources.",
     "dosage_growth_defect.parquet": "Gene dosage changes affecting growth.",
     "enamine_cloud_library_smiles.pkl": "Compounds from Enamine REAL library with SMILES annotations.",
     "evebio_assay_table.csv": "Assay metadata with one row per assay from EveBio pharmome mapping.",

biomni/env_desc_cm.py

@@ -19,7 +19,6 @@
     # "ddinter_hormonal.csv": "Drug-drug interactions for systemic hormonal preparations from DDInter 2.0 database.",  # CC BY-NC-SA 4.0 - Non-commercial only
     # "ddinter_respiratory.csv": "Drug-drug interactions for respiratory system drugs from DDInter 2.0 database.",  # CC BY-NC-SA 4.0 - Non-commercial only
     # "ddinter_various.csv": "Drug-drug interactions for various drugs from DDInter 2.0 database.",  # CC BY-NC-SA 4.0 - Non-commercial only
-    # "DisGeNET.parquet": "Gene-disease associations from multiple sources.",  # CC BY-NC-SA 4.0 - Non-commercial only
     "dosage_growth_defect.parquet": "Gene dosage changes affecting growth.",
     # "enamine_cloud_library_smiles.pkl": "Compounds from Enamine REAL library with SMILES annotations.",  # Proprietary - Requires license
     # "evebio_assay_table.csv": "Assay metadata with one row per assay from EveBio pharmome mapping.",  # Proprietary - Requires permission

biomni/llm.py

@@ -35,7 +35,7 @@ def get_llm(
     # Use config values for any unspecified parameters
     if config is not None:
         if model is None:
-            model = config.llm_model
+            model = config.llm
         if temperature is None:
             temperature = config.temperature
         if source is None: