Elucidating the Design Space of Flow Matching for Cellular Microscopy
For researchers in computational biology and microscopy, this work provides a simplified, scalable recipe for flow-matching models that significantly outperforms prior methods.
This paper analyzes the design space of flow-matching models for cellular microscopy, finding that many popular techniques are unnecessary. Their scaled model achieves a two-fold FID and ten-fold KID improvement over prior methods, and fine-tuning with molecular embeddings yields state-of-the-art performance for simulating responses to unseen molecules.
Flow-matching generative models are increasingly used to simulate cell responses to biological perturbations. However, the design space for building such models is large and underexplored. We systematically analyse the design space of flow matching models for cell-microscopy images, finding that many popular techniques are unnecessary and can even hurt performance. We develop a simple, stable, and scalable recipe which we use to train our foundation model. We scale our model to two orders of magnitude larger than prior methods, achieving a two-fold FID and ten-fold KID improvement over prior methods. We then fine-tune our model with pre-trained molecular embeddings to achieve state-of-the-art performance simulating responses to unseen molecules. Code is available at https://github.com/valence-labs/microscopy-flow-matching