Distance-AF improves predicted protein structure models by AlphaFold2 with user-specified distance constraints

Distance-AF 通过用户指定的距离约束,改进了 AlphaFold2 预测的蛋白质结构模型。

阅读:1

Abstract

The three-dimensional structure provides essential information for understanding biological functions of proteins. To aid structure determination, computational prediction has been extensively studied. Despite significant progress, challenges remain on difficult targets, such as those with multiple domains and proteins that fold into several conformations. Here we present Distance-AF, which aims to improve the performance of AlphaFold2 by incorporating distance constraints. Distance-AF reduced the root mean square deviation (RMSD) of structure models to native on average by 11.75 Å when compared to the models by AlphaFold2 on a test set of 25 targets. Distance-AF outperformed Rosetta and AlphaLink, which consider distance constraints. The average RMSD values for Distance-AF, Rosetta, and AlphaLink were 4.22 Å, 6.40 Å, and 14.29 Å, respectively. We further demonstrate its applications in various scenarios, including fitting structures into cryo-electron microscopy density maps, modeling active and inactive conformations, and generating conformational ensembles that satisfy Nuclear Magnetic Resonance data. Distance-AF has the potential to accelerate structural biology research, facilitate drug discovery, and provide a foundation for integrating experimental and computational approaches to study protein dynamics and interactions in complex biological systems.

特别声明

1、本页面内容包含部分的内容是基于公开信息的合理引用;引用内容仅为补充信息,不代表本站立场。

2、若认为本页面引用内容涉及侵权,请及时与本站联系,我们将第一时间处理。

3、其他媒体/个人如需使用本页面原创内容,需注明“来源:[生知库]”并获得授权;使用引用内容的,需自行联系原作者获得许可。

4、投稿及合作请联系:info@biocloudy.com。