PPIRef

PPIRef is a dataset of 3D structures of protein-protein interfaces. See the GitHub repository for more details.

File description

1. ppi_6A.zip stores the PPIRef dataset: .pdb files with all 6A-distance interfaces from PDB as of Jan 2024.
2. ppi_6A_stats.zip stores the statistics about the ppi_6A dataset. This includes the indexes for fast search with MMseqs2 and iDist, as well as a .csv file with main statistics for all interfaces.
3. ppi_10A.zip: .pdb files with all 10A-distance interfaces from PDB downloaded in June 2024.
4. ppi_10A_stats.zip stores the statistics about the ppi_10A dataset. This includes iDist embeddings, as well as a .csv file with main statistics for all interfaces.
5. pdb_redo_ppi_10A.zip: .pdb files with all 10A-distance interfaces from PDB-REDO downloaded in June 2024.
6. pdb_redo_ppi_10A_stats.zip stores the statistics about the pdb_redo_ppi_10A dataset. This includes iDist embeddings, as well as a .csv file with main statistics for all interfaces.
7. skempi2.zip stores PPI interfaces from the SKEMPI v2.0 dataset
8. benchmark_similarity_6A.zip: Results of benchmarking methods to calculate pairwise PPI similarity. The comparisons were performed on 2M 6A-interfaces. The directory contains results of iAlign and iDist. Please note that some values for iAlign are missing due to iAlign errors. Please see this GitHub issue for details of the benchmark construction.

How to use

It is recommended to download and extract the files in the PPIRef/ppiref/data/ppiref directory. This can be done automatically via the ppiref package. For example, to download and extract the ppi_6A.zip archive run:
from ppiref.utils.misc import download_from_zenodo
download_from_zenodo('ppi_6A.zip')