Accelerate drug discovery by predicting ADMET properties and safety risks for chemical compounds using advanced ML models.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install admet-prediction
2. Manual Installation
Copy the skill folder to one of these locations
~/.openclaw/skills/ <project>/skills/ Priority: Workspace > Local > Bundled
3. Prompt Installation
Copy this prompt to OpenClaw to install it automatically.
Help me install admet-prediction using Clawhub. If Clawhub is not installed, install it first (npm i -g clawhub).
Prefer to download?
Get the raw skill files in a ZIP archive.
What is ADMET Prediction?
The ADMET Prediction skill provides a robust framework for assessing the Absorption, Distribution, Metabolism, Excretion, and Toxicity of potential drug candidates. By leveraging QSAR and machine learning models, this addition to Openclaw Skills allows researchers to prioritize compounds based on drug-likeness, pharmacokinetic profiles, and early safety screens.
It automates the evaluation of critical factors such as Lipinski's Rule of Five, hERG inhibition, and metabolic clearance, significantly reducing the cost and time of early-stage pharmaceutical development. This skill is essential for any modern drug discovery pipeline looking to optimize compound selection before entering expensive in vivo trials.
ADMET Prediction Use Cases
- Early-stage screening of compound libraries to identify drug-like molecules.
- Assessing cardiotoxicity risk through hERG inhibition prediction.
- Evaluating metabolic stability and potential drug-drug interactions via CYP450 profiles.
- Predicting blood-brain barrier permeability for central nervous system targets.
- Filtering out Pan-Assay Interference Compounds to reduce false positives in assays.
How ADMET Prediction Works
- Input chemical structures via SMILES strings or SDF library files into the skill interface.
- The skill processes structures through specialized machine learning models for various pharmacokinetic properties.
- Calculations are performed for drug-likeness rules such as Veber, Egan, and Lipinski to ensure chemical viability.
- Toxicity alerts including DILI, Ames, and hERG are cross-referenced with structural alert databases.
- A comprehensive profile is generated, providing confidence scores and actionable recommendations for further testing.
ADMET Prediction Setup
Install the necessary chemoinformatics libraries using pip to get started with this component of Openclaw Skills:
pip install rdkit admet-models
# Optional for advanced deep learning models
pip install deepchem admet-x
ADMET Prediction Data Schema & Taxonomy
The skill organizes data into structured profiles and batch results. Output includes comprehensive tables and metadata taxonomy:
| Component | Data Type | Description |
|---|---|---|
| SMILES | String | The chemical structure representation input. |
| PK Properties | Float/Categorical | Quantitative predictions (e.g., Caco-2) or binary status. |
| Confidence | Percentage | Model certainty based on the applicability domain. |
| Recommendations | Text | Automated guidance on compound progression. |
Batch results are typically exported as .csv or .sdf files for downstream analysis within the Openclaw Skills ecosystem.
ADMET Prediction Advanced Features
- Batch library screening for large-scale compound collections in SDF format.
- Customizable rule sets for filtering molecules based on specific criteria like MDDR or Brenk alerts.
- Multi-model consensus scoring to significantly improve prediction reliability across diverse chemical spaces.
- Automated recommendation engine that flags specific structural alerts for medicinal chemistry optimization.
SKILL.md
Loading
Related Openclaw Skills
Price Monitor
An automated tool to track e-commerce prices and receive alerts for price drops or reached targets.
Markdown to Social Converter
A command-line utility that transforms markdown articles into platform-optimized social media posts for Twitter, LinkedIn, and Reddit.
Legal Docs FR
A Python-powered CLI generator for French legal documents, creating print-ready HTML for CGV, contracts, and quotes for freelancers.
French Services Skill
A comprehensive suite of Python scripts for accessing real-time French transport, logistics, and weather data.
Tavily Search
A high-performance web search skill optimized specifically for AI agents to retrieve clean, relevant real-time data.
Intent Guardian
An intelligent focus companion that monitors desktop activity to maintain a real-time task stack and remind you of forgotten work after interruptions.
Featured*
