2.6k
GitHub Stars
2
Bundled Files
2 months ago
Catalog Refreshed
4 months ago
First Indexed
Readme & install
Copy the install command, review bundled files from the catalogue, and read any extended description pulled from the listing source.
Installation
Preview and clipboard use veilstrat where the catalogue uses aiagentskills.
npx veilstrat add skill openclaw/skills --skill pharma-pharmacology-agent- _meta.json308 B
- SKILL.md5.1 KB
Overview
This skill profiles drug candidates from a SMILES string to produce ADME/PK, drug-likeness, and safety assessments. It computes Lipinski and Veber rules, QED, SA Score, PAINS alerts, and a set of rule-based ADME predictions. Output is a structured JSON report ready for downstream chaining to toxicology or IP workflows.
How this skill works
The agent accepts canonical SMILES and calculates RDKit descriptors (MW, logP, TPSA, HBD/HBA, rotatable bonds, aromatic rings). It applies rule-based heuristics and published models (ESOL, Egan, Clark-like rules) to predict BBB permeability, solubility, GI absorption, CYP3A4 inhibition risk, P-gp substrate likelihood, and plasma protein binding. The result is a validated JSON schema with flags, rationales, confidence scores, and recommended next steps.
When to use it
- Initial in-silico triage of virtual screening hits or design ideas
- Lead optimization to flag ADME liabilities early
- Prioritizing compounds for experimental ADME assays
- Automated pipelines that chain chemistry output to pharmacology profiling
- Batch profiling of chemical libraries before resource allocation
Best practices
- Provide canonical or validated SMILES to avoid parsing errors
- Use profiles to prioritize experimental follow-up, not as sole decision criteria
- Combine with in-vitro assays for metabolism and permeability validation
- Interpret rule-based risks (CYP3A4, P-gp, PPB) as triage flags requiring lab confirmation
- Chain output to toxicology and IP-expansion agents for end-to-end assessment
Example use cases
- Profile a single hit from a docking campaign to check drug-likeness and solubility
- Scan a small library to identify molecules with likely high GI absorption and low CYP3A4 risk
- Prioritize analogs during lead optimization by balancing QED, SA Score, and BBB prediction
- Reject or flag PAINS-containing chemotypes before investing in synthesis or assays
- Feed chemistry-query SMILES output into this agent for automated pipeline continuation
FAQ
Provide a valid SMILES string under the smiles field; the agent returns a clear error for invalid or missing input.
How reliable are the ADME predictions?
Predictions use validated heuristics and established approximations (ESOL, Egan, Clark-like rules). They are useful for triage but should be confirmed with experimental assays.
Can this handle batch processing?
Yes. The agent is designed to accept single SMILES per call and integrate into pipelines for batch processing via upstream chaining tools.