Setup & Installation
Or with the ClawHub CLI, for registry-managed skill folders outside a full OpenClaw workspace:
What This Skill Does
Predicts ADMET properties (Absorption, Distribution, Metabolism, Excretion, Toxicity) for drug candidates using ML/QSAR models. Takes molecular structures as SMILES or SDF files and returns property estimates with confidence scores. Covers drug-likeness rules, PK parameters, CYP interactions, and toxicity alerts including hERG and DILI.
Running predictions computationally before synthesis lets teams deprioritize compounds with poor PK or toxicity flags without spending lab resources on in vitro assays.
When to use it
- Screening a compound library for hERG cardiotoxicity risk before synthesis
- Checking oral bioavailability and Caco-2 permeability of a lead molecule
- Identifying CYP3A4 inhibition liability in a clinical candidate
- Filtering virtual compounds by Lipinski Ro5 and Veber rules
- Predicting BBB permeability for CNS-targeted drug candidates
Example Workflow
Here's how your AI assistant might use this skill in practice.
User asks: /admet "CC1=CC=C(C=C1)CNC" --full
- 1Parse SMILES input and validate molecular structure
- 2Run ML/QSAR models across all ADMET property categories
- 3Apply drug-likeness filters (Lipinski Ro5, Veber, PAINS, Brenk)
- 4Score toxicity endpoints including hERG, DILI, and Ames mutagenicity
- 5Compile property table with confidence levels and flag concerns
Full ADMET profile table with absorption, distribution, metabolism, excretion, and toxicity predictions, plus a recommendations section listing strengths, concerns, and a go/no-go suggestion.