UniRef #4
Description
The UniProt Reference Clusters (UniRef) are widely used for protein language model pre-training.
These include:
- UniRef100, composed of all UniProtKB records plus UniParc records not present in UniProt, available at this link.
- UniRef90, obtained by clustering UniRef100, available at this link.
- UniRef50, obtained by clustering UniRef90, available at this link.
The .fasta files need to be processed (?) and made easily available.