Compressing Biology: Evaluating the Stable Diffusion VAE for Phenotypic Drug Discovery

TL;DR

This paper systematically evaluates the Stable Diffusion VAE for reconstructing microscopy images in phenotypic drug discovery, demonstrating its effectiveness and comparing various evaluation metrics.

Contribution

It provides the first quantitative assessment of SD-VAE on microscopy data and benchmarks different evaluation methods for generative models in this domain.

Findings

SD-VAE preserves phenotypic signals with minimal loss

General-purpose feature extractors perform as well as bespoke models in retrieval tasks

Guidelines for evaluating generative models on microscopy data

Abstract

High-throughput phenotypic screens generate vast microscopy image datasets that push the limits of generative models due to their large dimensionality. Despite the growing popularity of general-purpose models trained on natural images for microscopy data analysis, their suitability in this domain has not been quantitatively demonstrated. We present the first systematic evaluation of Stable Diffusion's variational autoencoder (SD-VAE) for reconstructing Cell Painting images, assessing performance across a large dataset with diverse molecular perturbations and cell types. We find that SD-VAE reconstructions preserve phenotypic signals with minimal loss, supporting its use in microscopy workflows. To benchmark reconstruction quality, we compare pixel-level, embedding-based, latent-space, and retrieval-based metrics for a biologically informed evaluation. We show that general-purpose feature extractors like InceptionV3 match or surpass publicly available bespoke models in retrieval tasks, simplifying future pipelines. Our findings offer practical guidelines for evaluating generative models on microscopy data and support the use of off-the-shelf models in phenotypic drug discovery.