PolyConf: Unlocking Polymer Conformation Generation through Hierarchical Generative Models
Authors: Fanmeng Wang, Wentao Guo, Qi Ou, Hongshuai Wang, Haitao Lin, Hongteng Xu, Zhifeng Gao
ICML 2025 | Venue PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | The comprehensive evaluation demonstrates that Poly Conf consistently outperforms existing conformation generation methods, thus facilitating advancements in polymer modeling and simulation. The whole work is available at https: //polyconf-icml25.github.io. ... Extensive and comprehensive experiments on Poly Bench consistently demonstrate that our Poly Conf significantly outperforms existing conformation generation methods and achieves state-of-the-art performance |
| Researcher Affiliation | Collaboration | 1Gaoling School of Artificial Intelligence, Renmin University of China 2DP Technology 3California Institute of Technology 4SINOPEC Research Institute of Petroleum Processing Co., Ltd. 5Beijing Key Laboratory of Research on Large Models and Intelligent Governance 6Engineering Research Center of Next-Generation Intelligent Search and Recommendation, MOE. Correspondence to: Hongteng Xu <EMAIL>, Zhifeng Gao <EMAIL>. |
| Pseudocode | No | The paper describes the proposed method using prose and mathematical formulations, but it does not include any explicitly labeled pseudocode or algorithm blocks. |
| Open Source Code | Yes | The whole work, including code, model, and data, is publicly available to facilitate the development of this important yet largely unexplored research area. The whole work is available at https: //polyconf-icml25.github.io. |
| Open Datasets | Yes | we develop the first benchmark with a high-quality polymer conformation dataset derived from molecular dynamics simulations (Afzal et al., 2020) to boost related research in this area. ... The whole work, including code, model, and data, is publicly available to facilitate the development of this important yet largely unexplored research area. The whole work is available at https: //polyconf-icml25.github.io. |
| Dataset Splits | Yes | Specifically, the training set comprises about 46k polymers with their corresponding conformations, the validation set comprises about 5k polymers with their corresponding conformations, and the test set comprises about 2k polymers with their corresponding conformations. |
| Hardware Specification | No | The paper does not provide specific details about the hardware used for running the experiments, such as GPU/CPU models or other computational resources. |
| Software Dependencies | No | The paper mentions various software tools used, such as RDKit, Amber Tools (Antechamber, prepgen, TLeap), ACPYPE, and GROMACS, but does not specify their version numbers. |
| Experiment Setup | Yes | The optimization and MD simulations are conducted using GROMACS (Van Der Spoel et al., 2005) as the usage of GROMACS on polymer simulations has long been reported (Liu et al., 2024; Gr unewald et al., 2022). Specifically, the steepest descent method is applied to minimize the system energy, and then 5,000,000 steps (5ns) of MD calculations are performed at 298 K and 1 atm using the NVT ensemble for equilibrium calculations. |