Helical reconstruction in RELION is increasingly being used to determine the atomic structures of amyloid filaments from electron cryo-microscopy (cryo-EM) images. However, because the energy landscape of amyloid refinements is typically fraught with local optima, amyloid structure determination is often difficult. This paper aims to help RELION users in this process. It discusses aspects of helical reconstruction that are particularly relevant to amyloids, it illustrates the problem of local optima in refinement and how to detect them, and it introduces a new method to calculate 3D initial models from reference-free 2D class averages. By providing starting models that are closer to the global optimum, this method makes amyloid structure determination easier. All methods described are open-source and distributed within RELION-3.1. Their use is illustrated using a publicly available data set on tau filaments from the brain of an individual with Alzheimer's disease.

中文翻译：

---

RELION-3.1 中淀粉样蛋白结构测定。

RELION 中的螺旋重建越来越多地用于从电子冷冻显微镜 (cryo-EM) 图像中确定淀粉样蛋白丝的原子结构。然而，由于淀粉样蛋白细化的能量景观通常充满局部最优，因此淀粉样蛋白结构的确定通常很困难。本文旨在为 RELION 用户在此过程中提供帮助。它讨论了与淀粉样蛋白特别相关的螺旋重建的各个方面，说明了细化中的局部最优问题以及如何检测它们，并引入了一种根据无参考 2D 类平均值计算 3D 初始模型的新方法。通过提供更接近全局最优的起始模型，该方法使淀粉样蛋白结构的确定变得更容易。所描述的所有方法都是开源的并在 RELION-3.1 中分发。使用来自阿尔茨海默病患者大脑的 tau 蛋白丝的公开数据集说明了它们的用途。