Database introduction

Note：we have published embeddings of all these databases; download them from here. The Swiss-Prot database is on the HuggingFace page here.

Database	Introduction	Number of proteins	Reference link
Swiss-Prot	Human-reviewed protein sequence database	500K	https://www.uniprot.org/uniprotkb?query=reviewed:true
UniRef50	Generated by clustering UniProt proteins at 50% sequence identity	45M	https://www.uniprot.org/help/uniref
Uncharacterized	All proteins labeled as "Uncharacterized" at UniProt website	30M	https://www.uniprot.org/uniprotkb?query=Uncharacterized
OMG_prot50	Created by clustering the Open MetaGenomic dataset (OMG) at 50% sequence identity	200M	https://huggingface.co/datasets/tattabio/OMG_prot50
PDB	A database for the three-dimensional structural data of proteins	700K (every chain in a structure was extracted and counted as one protein)	https://www.rcsb.org
GOPC	Global ocean microbiome protein catalog sequences	2B	https://db.cngb.org/maya/datasets/MDB0000002
NCBI	NCBI protein database	700M	https://www.ncbi.nlm.nih.gov/protein
OMG	The Open MetaGenomic dataset (OMG)	3.1B	https://huggingface.co/datasets/tattabio/OMG
MGnify	MGnify is a free to use resource for analysis, visualisation and discovery of metagenomic, metatranscriptomic, amplicon and assembly datasets. We filtered out protein segments and only kept the full-length proteins.	470M	https://ftp.ebi.ac.uk/pub/databases/metagenomics/peptide_database/2024_04