CellFlux : Simulating Cellular Morphology Changes via Flow Matching
Yuhui Zhang, Yuchang Su, Chenyu Wang, Tianhong Li, Zoe Wefers, Jeffrey Nirschl, James Burgess, Daisy Ding, Alejandro Lozano, Emma Lundberg, Serena Yeung-Levy · 2025
Read CellFlux : Simulating Cellular Morphology Changes via Flow Matching by Yuhui Zhang, Yuchang Su, Chenyu Wang, Tianhong Li, Zoe Wefers, Jeffrey Nirschl, James Burgess, Daisy Ding, Alejandro Lozano, Emma Lundberg, Serena Yeung-Levy (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2502-09775.txt.
Introduction
Yuhui Zhang and a large team wrote a 2025 paper on CellFlux, a model that generates images of cells after chemical or genetic disturbance. A long-standing dream in computational biology is a virtual cell that predicts how real cells respond to perturbations. CellFlux moves toward it by learning how cell appearance transforms.
Using flow matching, the model learns the shift from unperturbed to perturbed cell populations as a whole, rather than copying single cells. That distribution view separates true drug effects from batch artifacts, a notorious confound. Tests covered chemical, genetic, and combined perturbation datasets. Generated images captured the specific shape changes of each treatment. Scores improved 35 percent on image quality and 12 percent on predicting a drug's mode of action. The model can also interpolate smoothly between states, tracing response dynamics.
Catalog record
| CREATOR | Yuhui Zhang, Yuchang Su, Chenyu Wang, Tianhong Li, Zoe Wefers, Jeffrey Nirschl, James Burgess, Daisy Ding, Alejandro Lozano, Emma Lundberg, Serena Yeung-Levy |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 62 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2502-09775
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
CellFlux : Simulating Cellular Morphology Changes via Flow Matching Yuhui Zhang, Yuchang Su, Chenyu Wang, Tianhong Li, Zoe Wefers, Jeffrey Nirschl, James Burgess, Daisy Ding, Alejandro Lozano, Emma Lundberg, Serena Yeung-Levy Abstract Building a virtual cell capable of accurately simulating cellular behaviors in silico has long been a dream in computational biology. We introduce CellFlux , an image-generative model that simulates cellular morphology changes induced by chemical and genetic perturbations using flow matching. Unlike prior methods, CellFlux models distribution-wise transformations from unperturbed to perturbed cell states, effectively distinguishing actual perturbation effects from experimental artifacts such as batch effects—a major challenge in biological data. Evaluated on chemical (BBBC021), genetic (RxRx1), and combined perturbation (JUMP) datasets, CellFlux generates biologically meaningful cell images that faithfully capture perturbation-specific morphological changes, achieving a 35% improvement in FID scores and a 12% increase in mode-of-action prediction accuracy over existing methods. Additionally, CellFlux enables continuous interpolation between cellular states, providing a potential tool for studying perturbation dynamics. These capabilities mark a significant step toward realizing virtual cell modeling for biomedical research. Project page: https://yuhui-zh15.github.io/CellFlux/ . † † affiliationnotice: Equal contribution 1 Introduction Figure 1: Overview of CellFlux . (a) Objective. CellFlux aims to predict changes in cell morphology induced by chemical or gene perturbations in silico . In this example, the perturbation effect reduces the nuclear size. (b) Data. The dataset includes images from high-content screening experiments, where chemical or genetic perturbations are applied to target wells, alongside control wells without perturbations. Control wells provide prior information to contrast with target images, enabling the identification of true perturbation effects (e.g., reduced nucleus size) while calibrating non-perturbation artifacts such as batch effects—systematic biases unrelated to the perturbation (e.g., variations in color intensity). (c) Problem formulation. We formulate the task as a distribution-to-distribution problem (many-to-many mapping), where the source distribution consists of control images, and the target distribution contains perturbed images within the same batch. (d) Flow matching. CellFlux employs flow matching, a state-of-the-art generative approach for distribution-to-distribution problems. It learns a neural network to approximate a velocity field, continuously transforming the source distribution into the target by solving an ordinary differential equation (ODE). (e) Results. CellFlux significantly outperforms baselines in image generation quality, achieving lower Fréchet Inception Distance (FID) and higher classification accuracy for mode-of-action (MoA) predictions. Building a virtual cell that simulates cellular behaviors in silico has been a longstanding dream in computational biology ( Slepchenko et al., 2003 ; Johnson et al., 2023 ; Bunne et al., 2024 ) . Such a system would revolutionize drug discovery by rapidly predicting how cells respond to new compounds or genetic modifications, significantly reducing the cost and time of biomedical research by prioritizing the experiments most likely to succeed based on the virtual cell simulation ( Carpenter, 2007 ) . Moreover, this could unlock personalized therapeutic development by building digital twins of cells from patients to simulate patient-specific responses ( Katsoulakis et al., 2024 ) . Two recent advances have made creating a generative virtual cell model possible. On the computational side, generative models now excel at modeling and sampling from complex data distributions, demonstrating remarkable success in synthesizing texts, images, videos, and biological sequences ( OpenAI, 2024 ; Esser et al., 2024 ; Kondratyuk et al., 2024 ; Hayes et al., 2025 ) . Concurrently, on the biotechnology side, automated high-content screening has generated massive imaging datasets — reaching terabytes or petabytes — that capture how cells respond to hundreds of thousands of chemical compounds and genetic modifications ( Chandrasekaran et al., 2023 ; Fay et al., 2023 ) . In this work, we introduce CellFlux , an image-generative model that simulates how cellular morphology changes in response to chemical or genetic perturbations (Figure 1 a). CellFlux ’s key innovation is formulating cellular morphology prediction as a distribution-to-distribution learning problem, and leveraging flow matching ( Lipman et al., 2023 ) , a state-of-the-art generative modeling technique designed for distribution-wise transformation, to solve this problem. Specifically, cell morphology data are collected through high-content microscopy screening, where images of control and perturbed cells are captured from experimental wells across different batches (Figure 1 b). Control wells, which receive no drug treatment or genetic modifications, play a crucial role in providing prior information and serving as a reference to distinguish true perturbation effects from other sources of variation. They help calibrate non-perturbation factors, such as batch effects—systematic biases unrelated to perturbations, including variations in color or intensity, akin to distribution shifts in machine learning. Properly incorporating control wells is essential for capturing actual perturbation effects rather than artifacts, yet many existing methods overlook this aspect ( Yang et al., 2021 ; Navidi et al., 2025 ; Cook et al., 2024 ) . To address this, we frame cellular morphology prediction as a distribution-to-distribution mapping problem (Figure 1 c), where the source distribution consists of control cell images, and the target distribution comprises perturbed cell images from the same batch. To address this distribution-to-distribution problem, CellFlux employs flow matching, a state-of-the-art generative modeling approach designed for distribution-wise transformations (Figure 1 d). The framework continuously transforms the source distribution into the target using an ordinary differential equation (ODE) by learning a neural network to approximate a velocity field. This direct and native distribution transformation enabled by flow matching is intuitively more effective than previous methods, which rely on adding extra components to GANs, incorporating the source as a condition, or mapping between distributions and noise using diffusion models ( Palma et al., 2025 ; Hung et al., 2024 ; Bourou et al., 2024 ) . We demonstrate the effectiveness of CellFlux on three datasets: BBBC021 (chemical perturbations) ( Caie et al., 2010 ) , RxRx1 (genetic modifications via CRISPR or ORF) ( Sypetkowski et al., 2023 ) , and JUMP (combined chemical and genetic perturbations) ( Chandrasekaran et al., 2023 ) . CellFlux generates high-fidelity images of cellular changes in response to perturbations across all datasets, improving FID scores by 35% over previous approaches. The generated images capture meaningful biological patterns, demonstrated by a 12% improvement in predicting mode-of-action compared to existing methods (Figure 1 e). Importantly, CellFlux maintains consistent performance across diverse experimental conditions and generalizes to held-out perturbations never seen during training, showing its broad applicability. Moreover, CellFlux introduces two key capabilities with significant potential for biological research (Figure 4 ). First, it effectively corrects batch effects by conditioning on control cells from different batches. By comparing control images with generated images, it can disentangle true perturbation-induced morphological changes from experimental batch artifacts. Second, CellFlux enables bidirectional interpolation between cellular states due to the continuous and reversible nature of the velocity field in flow matching. This interpolation provides a means to explore intermediate cellular morphologies and potentially gain deeper insights into dynamic perturbation responses. In summary, by formulating cellular morphology prediction as a distribution-to-distribution problem and using flow matching as a solution, CellFlux enables accurate prediction of perturbation responses (Figure 1 ). CellFlux not only achieves state-of-the-art performance but unlocks new capabilities such as handling batch effects or visualizing cellular state transitions, significantly advancing the field towards a virtual cell for drug discovery and personalized therapy. 2 Problem Formulation In this section, we introduce the objective, data, and mathematical formulation of