What problem does it solve?
Manually sorting CRISPR screen fold changes ignores follow-up feasibility and produces inconsistent hit rankings. This Skill combines guide depletion, essentiality, and druggability into one transparent, reproducible triage score computed entirely locally.
Core Features & Use Cases
- Guide-to-gene aggregation: Computes guide-level log2 fold change from control and treatment counts, then collapses guides per gene using the median so one noisy guide cannot dominate.
- Deterministic triage scoring: Scores each gene with fixed weights (0.55 depletion, 0.25 druggability, 0.20 essentiality) and assigns high, medium, or watch priority tiers.
- Reproducible report pack: Writes a ranked Markdown report, machine-readable JSON, gene and guide CSV tables, and a commands.sh audit trail.
- Use Case: A functional genomics researcher finishes a knockout screen and needs to decide which depleted genes to follow up. They run the skill on their guide count CSV and get BRCA1-style ranked hits with priority tiers in seconds.
Quick Start
Run the CRISPR screen triage demo with "python clawbio.py run crispr-triage --demo" or pass your own guide count CSV via the --input flag to generate a ranked hit report.
