Chem2Gen-Bench: Benchmarking Chemical-to-Genetic Translation in Perturbation Response Space

2026-06-19

AI Summary

This study addresses the lack of a systematic evaluation framework for assessing the interchangeability of chemical and genetic perturbations in target-matching contexts. To bridge this gap, the authors introduce Chem2Gen-Bench, a benchmark integrating 260,000 chemical and 1.09 million genetic perturbation profiles organized by cell–target context—the first evaluation suite specifically designed for perturbation translation. The framework systematically evaluates foundation models through perturbation alignment, similarity retrieval, covariate modeling, and representation space analysis, augmented by auditable alignment verification and background correction strategies. Key findings reveal heterogeneous fidelity in perturbation translation; while background correction strengthens the association between retrieval success and similarity, it reduces average performance. Moreover, current model embeddings do not consistently surpass a simple gene-difference baseline.

Abstract

Virtual-cell and perturbation models are increasingly used to predict cellular responses for biomedical discovery, but chemical and genetic perturbations are not automatically interchangeable. Existing evaluations often study chemical response prediction or genetic perturbation prediction separately, leaving target-matched chemical-to-genetic translation under-tested. We introduce Chem2Gen-Bench, a benchmark comprising 260,084 chemical and 1,099,045 genetic perturbation profiles organized into cell-target contexts, and evaluate pairwise alignment, retrieval, protocol covariate associations, feature spaces, and foundation-model embeddings. Across matched contexts, translation fidelity is measurable but heterogeneous; background adjustment increases the association between pairwise similarity and retrieval success, while paired tests show lower mean retrieval success after adjustment under the evaluated settings. In a target-matched K562 audit, the evaluated foundation-model embeddings did not consistently improve over gene-delta baselines. Chem2Gen-Bench provides an auditable framework for testing when chemical and genetic perturbations align around shared targets and when representation gains are supported by matched perturbation evidence.

Problem

• chemical-to-genetic translation
• perturbation response
• target-matched perturbations
• benchmarking
• virtual-cell models

Innovation

• Chem2Gen-Bench
• chemical-to-genetic translation
• perturbation response
• foundation-model embeddings
• target-matched benchmarking