# P2DFlow: A Protein Ensemble Generative Model with SE(3) Flow Matching

arXiv: https://arxiv.org/abs/2411.17196 · HTML: https://mlsanity.com/paper/2411.17196/

## Key facts

- Cited by 6 papers in the map (data through 14 Oct 2026).
- Builds on 0 of its 15 references that are in the map.
- First posted on arXiv 26 Nov 2024.

## Built on by (most cited first, 6 of 6)

- [Flow Matching Meets Biology and Life Science: A Survey](https://mlsanity.com/paper/2507.17731/) (2025)
- [Generative Modeling of Full-Atom Protein Conformations using Latent Diffusion on Graph Embeddings](https://mlsanity.com/paper/2506.17064/) (2025)
- [Flash Invariant Point Attention](https://mlsanity.com/paper/2505.11580/) (2025)
- [AI-based Methods for Simulating, Sampling, and Predicting Protein Ensembles](https://mlsanity.com/paper/2509.17224/) (2025)
- [Learning Structure, Energy, and Dynamics: A Survey of Artificial Intelligence for Protein Dynamics](https://mlsanity.com/paper/2604.25244/) (2026)
- [GeoGraph: Geometric and Graph-based Ensemble Descriptors for Intrinsically Disordered Proteins](https://mlsanity.com/paper/2510.00774/) (2025)

## Abstract (arXiv)

Biological processes, functions, and properties are intricately linked to the ensemble of protein conformations, rather than being solely determined by a single stable conformation. In this study, we have developed P2DFlow, a generative model based on SE(3) flow matching, to predict the structural ensembles of proteins. We specifically designed a valuable prior for the flow process and enhanced the model's ability to distinguish each intermediate state by incorporating an additional dimension to describe the ensemble data, which can reflect the physical laws governing the distribution of ensembles, so that the prior knowledge can effectively guide the generation process. When trained and evaluated on the MD datasets of ATLAS, P2DFlow outperforms other baseline models on extensive experiments, successfully capturing the observable dynamic fluctuations as evidenced in crystal structure and MD simulations. As a potential proxy agent for protein molecular simulation, the high-quality ensembles generated by P2DFlow could significantly aid in understanding protein functions across various scenarios. Code is available at https://github.com/BLEACH366/P2DFlow