ChemCrow Tools
ChemCrow Tools provides a comprehensive suite of functions for drug discovery agents. It wraps RDKit (with safe fallbacks) to calculate properties, check Lipinski rules, and screen for toxicity.
When to Use This Skill
- When evaluating a candidate molecule for drug-likeness (Lipinski Rule of 5).
- To check for toxic substructures (PAINS, Ames mutagenicity alerts).
- To calculate molecular weight, LogP, and TPSA.
- To estimate synthetic accessibility (SA Score).
Core Capabilities
- Property Calculation: MolWt, LogP, TPSA, QED.
- Safety Screening: Detects toxicophores and PAINS.
- Synthesizability: Estimates how hard a molecule is to make.
Workflow
- Input: Provide a valid SMILES string.
- Analyze: The tool runs a battery of RDKit descriptors and SMARTS pattern matching.
- Output: Returns a JSON report with properties and a recommendation.
Example Usage
User: "Analyze the safety and properties of Aspirin."
Agent Action:
python3 Skills/Drug_Discovery/ChemCrow_Tools/chem_tools.py \
--smiles "CC(=O)OC1=CC=CC=C1C(=O)O" \
--output aspirin_analysis.json