Tdc Oracle Tools¶
Configuration File: tdc_oracle_tools.json
Tool Type: Local
Tools Count: 1
This page contains all tools defined in the tdc_oracle_tools.json configuration file.
Available Tools¶
TDC_predict_oracle_score (Type: TDCOracleTool)¶
Score one or more molecules (SMILES) with a pretrained Therapeutics Data Commons (TDC) oracle, co…
TDC_predict_oracle_score tool specification
Tool Information:
Name:
TDC_predict_oracle_scoreType:
TDCOracleToolDescription: Score one or more molecules (SMILES) with a pretrained Therapeutics Data Commons (TDC) oracle, computed locally via the PyTDC package. Oracles are ML/medicinal-chemistry scorers used for molecular property prediction and goal-directed optimization. Supported oracles: ‘QED’ (drug-likeness 0-1, higher=better; RDKit, offline), ‘SA’ (synthetic accessibility 1=easy to 10=hard; downloads a small table once), ‘LogP’ (TDC’s PENALIZED logP = Crippen logP - synthetic-accessibility - ring penalty, an optimization objective, NOT raw octanol-water logP; for plain logP use RDKit/SwissADME; RDKit, offline), ‘GSK3B’ (ML probability of GSK3-beta inhibition 0-1; downloads a model once), ‘JNK3’ (ML probability of JNK3 inhibition 0-1; downloads a model once), ‘DRD2’ (ML probability of DRD2 activity 0-1; downloads a model once). Distinct from RDKit/ADMET tools because it exposes TDC’s pretrained bioactivity oracles. Requires the optional ‘PyTDC’ package; returns a clean error if it is not installed.
Parameters:
smiles(unknown) (required) A single SMILES string or a list of SMILES strings to score. Example: ‘CC(=O)Oc1ccccc1C(=O)O’ (aspirin).oracle(string) (required) Oracle name (case-insensitive). One of: QED, SA, LogP, GSK3B, JNK3, DRD2.
Example Usage:
query = {
"name": "TDC_predict_oracle_score",
"arguments": {
"smiles": "example_value",
"oracle": "example_value"
}
}
result = tu.run(query)