ATOM3D

ATOM3D 是关于生物分子三维结构的数据集的统一集合，包括蛋白质、小分子和核酸。这些数据集专门设计用于为在 3D 分子结构上运行的机器学习方法提供基准，并代表各种重要的结构、功能和工程任务。所有数据集都以标准化格式提供，以及一个 Python 包，其中包含用于常见机器学习框架（如 PyTorch）的处理代码、实用程序、模型和数据加载器。 ATOM3D 被设计为一个动态数据库，其中数据集会随着领域的进展而更新并添加任务。
ATOM3D 目前包含八个数据集，大致可分为四类，代表了广泛的问题，涵盖单个分子结构和生物分子之间的相互作用以及分子功能和设计/工程任务。

ATOM3D is a unified collection of datasets focused on the three-dimensional (3D) structures of biological molecules, including proteins, small molecules, and nucleic acids. These datasets are specifically designed to provide benchmark resources for machine learning methods that operate on 3D molecular structures, covering a variety of critical structural, functional, and engineering tasks. All datasets are provided in a standardized format, alongside a Python package that includes processing code, utilities, models, and data loaders for common machine learning frameworks such as PyTorch. ATOM3D is constructed as a dynamic database, where datasets and tasks will be updated and added as the field progresses. Currently, ATOM3D contains eight datasets, which can be roughly categorized into four categories, representing a wide range of problems spanning individual molecular structures, biomolecular interactions, molecular function, as well as molecular design and engineering tasks.

搜集汇总

背景与挑战

背景概述

ATOM3D是一个生物分子三维结构数据集的统一集合，包括蛋白质、小分子和核酸，专门用于为基于3D结构的机器学习方法提供基准。该数据集以标准化格式提供，并包含处理代码和实用程序，涵盖多种结构、功能和工程任务，且会动态更新以反映领域进展。