Chris-young-2004/PepSpecBench

PepSpecBench是一个专为肽段MS/MS谱预测设计的统一基准数据集。数据集提供了两种不同的表示形式：native（完整的全谱表示）和canonical（标准化的234维空间表示）。包含三个主要数据组：massive_kb（来自MassIVE-KB的数据表）、prospect（来自PROSPECT数据集的数据表）和ood（跨物种分布外评估表）。此外，还提供了一个轻量级的样本数据包sample_data用于快速检查。

PepSpecBench is a unified benchmark dataset designed for peptide MS/MS spectrum prediction. The data is provided in two distinct representations: native (the complete, benchmark-ready full-spectrum representation) and canonical (a processed representation in a standardized 234-dimensional space). The dataset is distributed across three primary data groups: massive_kb (benchmark-ready data tables derived from MassIVE-KB), prospect (benchmark-ready data tables derived from the PROSPECT dataset), and ood (cross-species out-of-distribution evaluation tables). A lightweight sample package is also provided in the sample_data config for rapid preliminary inspection.