Sequence-structure-function relationships in the microbial protein universe

The Microbiome Immunity Project (MIP) dataset contains models predicted with both Rosetta and DMPFold (folder `dataset/`). It also contains DeepFRI function predictions for all models. The `metadata` folder contains additional data which may be useful for searching the MIP database (FASTA files, BLAST databases and useful scripts for structure/function search) as well as retrieving the sequence/structural annotations. The `intermediate_data` folder contains preprocessed output for reproducing many of the figures in our manuscript in conjunction with scripts and Juypter notebooks found in our git repository: https://github.com/microbiome-immunity-project/protein_universe . More information about the dataset and associated metadata is provided in the `README.md` file). We are also providing workflows to search the MIP database against a protein sequence or structure or function of interest (see `SEARCHING.md` for more details).