What problem does it solve?
Deciding whether a drug target is worth pursuing requires synthesizing genetic, structural, chemical, safety, and clinical evidence scattered across dozens of databases. This Skill automates that multi-dimensional assessment and produces a quantitative Target Validation Score with a clear GO/NO-GO recommendation before committing to wet-lab work.
Core Features & Use Cases
- 10-Phase Validation Pipeline: Runs target disambiguation, disease association, druggability, chemical matter, clinical precedent, safety, pathway context, validation evidence, structural insights, and literature analysis using 60+ ToolUniverse tools (OpenTargets, ChEMBL, PDB, GTEx, GWAS, DepMap, and more).
- Quantitative Scoring & Evidence Grading: Computes a 0-100 composite score across five dimensions, assigns priority tiers (Tier 1 GO through Tier 4 NO-GO), and grades every claim as T1 (clinical proof) through T4 (computational prediction).
- ML Predictor Integration: Mandatorily runs ADMET-AI, AlphaFold, ESMFold, and DoGSite deep-learning models alongside database lookups, with a head-to-head ADMET comparison table for candidate drugs.
- Use Case: Ask "Is KRAS a druggable target for pancreatic cancer?" and receive a full markdown validation report with scorecard, risk assessment, recommended experiments, tool compounds, and biomarker strategy.
Quick Start
Ask the agent to validate whether a specific gene is a good drug target for a given disease, for example by requesting a full target validation of EGFR for non-small cell lung cancer.
