ALTAS MD trajectories

该数据集包含三条平行的轨迹，每条蛋白质轨迹有10,001帧，用于训练AlphaFlow-Lit模型。此外，为了创建训练集，这些轨迹以每100帧的步长进行下采样，总计300帧。数据集的规模分为1265个训练集、39个验证集和82个测试集，任务旨在生成蛋白质集合。

This dataset contains three parallel trajectories, each consisting of 10,001 frames, for training the AlphaFlow-Lit model. Furthermore, to create the training set, these trajectories were downsampled with a stride of every 100 frames, yielding a total of 300 frames per trajectory. The dataset is partitioned into 1265 training samples, 39 validation samples, and 82 test samples. The designated task for this dataset is to generate protein ensembles.