|国家预印本平台
首页|PFMBench: Protein Foundation Model Benchmark

PFMBench: Protein Foundation Model Benchmark

PFMBench: Protein Foundation Model Benchmark

来源:Arxiv_logoArxiv
英文摘要

This study investigates the current landscape and future directions of protein foundation model research. While recent advancements have transformed protein science and engineering, the field lacks a comprehensive benchmark for fair evaluation and in-depth understanding. Since ESM-1B, numerous protein foundation models have emerged, each with unique datasets and methodologies. However, evaluations often focus on limited tasks tailored to specific models, hindering insights into broader generalization and limitations. Specifically, researchers struggle to understand the relationships between tasks, assess how well current models perform across them, and determine the criteria in developing new foundation models. To fill this gap, we present PFMBench, a comprehensive benchmark evaluating protein foundation models across 38 tasks spanning 8 key areas of protein science. Through hundreds of experiments on 17 state-of-the-art models across 38 tasks, PFMBench reveals the inherent correlations between tasks, identifies top-performing models, and provides a streamlined evaluation protocol. Code is available at \href{https://github.com/biomap-research/PFMBench}{\textcolor{blue}{GitHub}}.

Mengdi Liu、Zhangyang Gao、Cheng Tan、Chenrui Xu、Hao Wang、Xiaoming Zhang、Stan Z. Li、Bozhen Hu、Linlin Chao

生物科学现状、生物科学发展生物科学研究方法、生物科学研究技术生物工程学计算技术、计算机技术

Mengdi Liu,Zhangyang Gao,Cheng Tan,Chenrui Xu,Hao Wang,Xiaoming Zhang,Stan Z. Li,Bozhen Hu,Linlin Chao.PFMBench: Protein Foundation Model Benchmark[EB/OL].(2025-06-01)[2025-06-30].https://arxiv.org/abs/2506.14796.点此复制

评论