PDF(Pubmed)
Abstract：
When solving a structure of a protein from single-wavelength anomalous diffraction X-ray data, the initial phases obtained by phasing from an anomalously scattering substructure usually need to be improved by an iterated electron-density modification. In this manuscript, the use of convolutional neural networks (CNNs) for segmentation of the initial experimental phasing electron-density maps is proposed. The results reported demonstrate that a CNN with U-net architecture, trained on several thousands of electron-density maps generated mainly using X-ray data from the Protein Data Bank in a supervised learning, can improve current density-modification methods.
摘要：
从单波长异常衍射X射线数据求解蛋白质的结构时，通过从异常散射子结构定相获得的初始相通常需要通过迭代电子密度修改来改善。在这份手稿中,提出了使用卷积神经网络（CNN）分割初始实验定相电子密度图。报告的结果表明，具有U网架构的CNN，在监督学习中，主要使用蛋白质数据库中的X射线数据生成的数千个电子密度图进行训练，可以提高电流密度的改性方法。
