Gespeichert in:
Bibliographische Detailangaben
Hauptverfasser: Wu, Dongxia, Su, Shiye, Zhang, Yuhui, Sui, Elaine, Lundberg, Emma, Fox, Emily B., Yeung-Levy, Serena
Format: Preprint
Veröffentlicht: 2026
Schlagworte:
Online-Zugang:https://arxiv.org/abs/2603.21743
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
_version_ 1866918515502481408
author Wu, Dongxia
Su, Shiye
Zhang, Yuhui
Sui, Elaine
Lundberg, Emma
Fox, Emily B.
Yeung-Levy, Serena
author_facet Wu, Dongxia
Su, Shiye
Zhang, Yuhui
Sui, Elaine
Lundberg, Emma
Fox, Emily B.
Yeung-Levy, Serena
contents 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.
format Preprint
id arxiv_https___arxiv_org_abs_2603_21743
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle CellFluxRL: Biologically-Constrained Virtual Cell Modeling via Reinforcement Learning
Wu, Dongxia
Su, Shiye
Zhang, Yuhui
Sui, Elaine
Lundberg, Emma
Fox, Emily B.
Yeung-Levy, Serena
Machine Learning
Quantitative Methods
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.
title CellFluxRL: Biologically-Constrained Virtual Cell Modeling via Reinforcement Learning
topic Machine Learning
Quantitative Methods
url https://arxiv.org/abs/2603.21743