Many Needles in a Haystack: Active Hit Discovery for Perturbation Experiments

TL;DR

This paper introduces a Bayesian optimization-based method called Probability-of-Hit for efficient hit discovery in gene perturbation experiments, outperforming traditional approaches by focusing on threshold exceedance.

Contribution

It formalizes hit discovery as a sequential design problem and proposes a novel acquisition function with proven asymptotic optimality.

Findings

Up to 6.4% improvement over baselines on biological datasets

Strong empirical performance on synthetic and real data

Asymptotic optimality of the proposed method

Abstract

High-throughput gene perturbation experiments can test several genetic interventions in parallel, yet experimental budgets remain limited. A central goal is hit discovery: identifying as many perturbations as possible whose phenotypic effect exceeds a predefined threshold. Pure exploration strategies are statistically inefficient, wasting budget on low-value regions. Bayesian optimization methods offer a principled alternative but target a single global optimum, over-exploiting dominant modes while neglecting other high-value regions. We formalize hit discovery as a sequential experimental design problem and propose Probability-of-Hit, an acquisition function that directly targets threshold exceedance by ranking candidates according to their posterior probability of being a hit. We prove asymptotic optimality of this approach and demonstrate strong empirical performance on both synthetic benchmarks and real biological immunology datasets, including up to 6.4% improvement over baselines on the Schmidt IL-2 dataset.