Institute for Nanostructure and Solid State Physics, Universität Hamburg, Luruper Chaussee 149 / Bldg. 610 (HARBOR), 22761 Hamburg, Germany.
Curr Opin Struct Biol. 2022 Jun;74:102368. doi: 10.1016/j.sbi.2022.102368. Epub 2022 Apr 15.
Machine learning methods, in particular convolutional neural networks, have been applied to a variety of problems in cryo-EM and macromolecular crystallographic structure solution. However, they still have only limited acceptance by the community, mainly in areas where they replace repetitive work and allow for easy visual checking, such as particle picking, crystal centering or crystal recognition. With Artificial Intelligence (AI) based protein fold prediction currently revolutionizing the field, it is clear that their scope could be much wider. However, whether we will be able to exploit this potential fully will depend on the manner in which we use machine learning: training data must be well-formulated, methods need to utilize appropriate architectures, and outputs must be critically assessed, which may even require explaining AI decisions.
机器学习方法，特别是卷积神经网络，已被应用于 cryo-EM 和大分子晶体学结构解析的各种问题。然而，它们仍然只被社区有限地接受，主要是在它们可以替代重复工作并允许轻松进行可视化检查的领域，例如粒子挑选、晶体中心定位或晶体识别。基于人工智能（AI）的蛋白质折叠预测目前正在彻底改变这个领域，很明显，它们的应用范围可能会更广。然而，我们是否能够充分利用这种潜力将取决于我们使用机器学习的方式：训练数据必须精心设计，方法需要利用适当的架构，输出必须进行严格评估，这甚至可能需要解释 AI 决策。
