Biology-in-the-loop: Amortized Adaptive Hit Discovery in CRISPR Screens
The paper claims an adaptive CRISPR-screening system can find far more hits while testing only a small slice of candidates.
The authors introduce AssayBench-Loop, a benchmark of 1,389 CRISPR screens across five phenotype categories. Their AssayLoop framework trains an acquisition policy on historical screens, then combines that learned strategy with LLM-derived biological priors. On temporally held-out screens, it reports 5.67x enrichment over random selection and recovers 27.7% of hits after assaying about 5% of the library. ArXiv · AI/CL/LG's note
The authors introduce AssayBench-Loop, a benchmark of 1,389 CRISPR screens across five phenotype categories. Their AssayLoop framework trains an acquisition policy on historical screens, then combines that learned strategy with LLM-derived biological priors. On temporally held-out screens, it reports 5.67x enrichment over random selection and recovers 27.7% of hits after assaying about 5% of the library. ArXiv · AI/CL/LG's note
score 5