CellFluxRL: Biologically-Constrained Virtual Cell Modeling via Reinforcement Learning

TL;DR

This paper introduces CellFluxRL, a reinforcement learning framework that enhances virtual cell models by enforcing biological and physical constraints, leading to more meaningful cellular simulations for drug discovery.

Contribution

It presents a novel RL-based post-training method that improves biologically plausible cell image generation over existing generative models.

Findings

CellFluxRL outperforms the original CellFlux model across all biological and structural rewards.

Test-time scaling further enhances the model's performance.

The framework advances virtual cell modeling towards biologically meaningful simulations.

Abstract

Building virtual cells with generative models to simulate cellular behavior in silico is emerging as a promising paradigm for accelerating drug discovery. However, prior image-based generative approaches can produce implausible cell images that violate basic physical and biological constraints. To address this, we propose to post-train virtual cell models with reinforcement learning (RL), leveraging biologically meaningful evaluators as reward functions. We design seven rewards spanning three categories-biological function, structural validity, and morphological correctness-and optimize the state-of-the-art CellFlux model to yield CellFluxRL. CellFluxRL consistently improves over CellFlux across all rewards, with further performance boosts from test-time scaling. Overall, our results present a virtual cell modeling framework that enforces physically-based constraints through RL, advancing beyond "visually realistic" generations towards "biologically meaningful" ones.