category
PNAS
date
Mar 4, 2026
slug
status
Published
summary
揭示AlphaFold3预测中存在因训练数据量不足导致的系统性偏差，指出配体结合状态的预测准确性更多取决于训练数据规模而非配体本身的存在。
tags
蛋白质组学
type
Post
📄 原文题目
Bias in the AlphaFold3 prediction of ligand-induced domain motion in enzymes
🔗 原文链接
💡 AI 核心解读
揭示AlphaFold3预测中存在因训练数据量不足导致的系统性偏差，指出配体结合状态的预测准确性更多取决于训练数据规模而非配体本身的存在。
📝 英文原版摘要
Proceedings of the National Academy of Sciences, Volume 123, Issue 10, March 2026. <br />SignificanceCofolding methods such as AlphaFold 3 (AF3) enable the prediction of protein structures with and without bound ligands. However, results may depend more on the number of structures available for training than on the presence of a ligand in the ...
- 作者:NotionNext
- 链接:https://tangly1024.com/article/32248bd6-1f96-81bf-b6b4-f22d4cc16b2f
- 声明:本文采用 CC BY-NC-SA 4.0 许可协议，转载请注明出处。
相关文章