RosettaCommons/MIP

该数据集名为Microbiome Immunity Project: Protein Universe，主要包含来自微生物生命树中1,003个代表性基因组的约200,000个预测蛋白质结构，并对这些结构进行了基于残基的功能注释。数据集的生成过程包括从GEBA1003参考基因组数据库中提取非冗余的蛋白质序列，使用Rosetta和DMPfold模型进行大规模结构预测，并通过DeepFRI进行功能注释。数据集分为高质量和低质量模型，并提供了详细的配置信息和下载链接。

The Microbiome Immunity Project: Protein Universe dataset contains ~200,000 predicted protein structures from 1,003 representative genomes across the microbial tree of life, annotated functionally on a per-residue basis. These structures were generated using Rosetta and DMPfold models, and the dataset includes both high-quality and low-quality models. Functional annotations were created using DeepFRIs structure-based Graph Convolutional Network embeddings. The dataset is part of the Genomic Encyclopedia of Bacteria and Archaea (GEBA1003) project, focusing on protein domains from non-redundant gene catalogs. The dataset is licensed under CC-BY-4.0 and is available for exploration of sequence-structure-function relationships in microbial proteins.