metadata
license: cc0-1.0
size_categories:
- 1M<n<10M
This is a parquet-ified subset of the RNAcentral active fasta file available here: https://ftp.ebi.ac.uk/pub/databases/RNAcentral/releases/24.0/sequences/rnacentral_active.fasta.gz
I have preprocessed it a bit, requiring only sequences less than 8192 nt long, and having no ambiguous nucleotides (i.e. no Ns or other non standard things)
This dataset is about 10% of the overall, and comprises 3,252,483 (3.2 million) sequences, or 2,642,703,990 (2.6 billion) bases.
The train/val/test split is 60/20/20