Notice: The reproducibility variables underlying each score are classified using an automated LLM-based pipeline, validated against a manually labeled dataset. LLM-based classification introduces uncertainty and potential bias; scores should be interpreted as estimates. Full accuracy metrics and methodology are described in Coakley et alK. L. Coakley, T. Snelleman, H. Hoos, and O. E. Gundersen, "The embrace of open science: An analysis of a decade of AI research and 56 800 conference papers," Under Review, 2026..
Modeling Cell Dynamics and Interactions with Unbalanced Mean Field Schrödinger Bridge
Authors: Zhenyi Zhang, Zihan Wang, Yuhao Sun, Tiejun Li, Peijie Zhou
NeurIPS 2025 | Venue PDF | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | The effectiveness of our method has been extensively validated using both synthetic gene regulatory data and real sc RNA-seq datasets. Compared to existing methods, Cyto Bridge identifies growth, transition, and interaction patterns, eliminates false transitions, and reconstructs the developmental landscape with greater accuracy. |
| Researcher Affiliation | Academia | 1LMAM and School of Mathematical Sciences, Peking University. 2Center for Quantitative Biology, Peking University. 3Center for Machine Learning Research, Peking University. 4NELBDA, Peking University. 5AI for Science Institute, Beijing. Emails: EMAIL, EMAIL |
| Pseudocode | Yes | Algorithm 1 Simulating ODEs: RBM |
| Open Source Code | Yes | Code is available at: https://github.com/zhenyiizhang/Cyto Bridge-Neur IPS. |
| Open Datasets | Yes | The effectiveness of our method has been extensively validated using both synthetic gene regulatory data and real sc RNA-seq datasets. Specifically, for the modeling of interaction potential, we set the number of RBF kernels to 8 across different datasets. For real-world sc RNA-seq data, the threshold dcutoff is set no lower than the largest distance between cells in specific datasets so that all pairs of cells are involved in interacting with each other. For the zebrafish data, we set the threshold dcutoff in the physical space, where cells interact with neighboring cells in physical space. We conducted experiments on the effect of different values of dcutoff in Table 16. |
| Dataset Splits | Yes | For all datasets, models were trained using all available time points and were evaluated using W1 and TMV. For synthetic gene regulatory networks with attractive interactions, mouse hematopoiesis, and embryoid body dataset, additional experiments with one-time point held out were conducted. |
| Hardware Specification | Yes | The experiments were performed on a shared high-performance computing cluster with NVIDIA A100 GPU and 128 CPU cores. |
| Software Dependencies | No | The paper mentions 'Pytorch (Paszke et al. 2017)' but does not specify a version number for PyTorch or any other software. |
| Experiment Setup | Yes | A Training Details |