---
dataset_info:
  features:
  - name: Entry
    dtype: string
  - name: classes
    dtype: string
  - name: Sequence
    dtype: string
  - name: labels
    dtype: int64
  splits:
  - name: train
    num_bytes: 102966725
    num_examples: 252729
  - name: valid
    num_bytes: 2035250
    num_examples: 5000
  - name: test
    num_bytes: 2022479
    num_examples: 5000
  download_size: 103184837
  dataset_size: 107024454
configs:
- config_name: default
  data_files:
  - split: train
    path: data/train-*
  - split: valid
    path: data/valid-*
  - split: test
    path: data/test-*
---
UniProt search (7/22/2025): * (Swiss-Prot reviewed entries only) download TSV with Taxonomic lineage (Ids), sequence, and length.

Curation steps:

- Filtered for sequence length >=20 and <=2048
- Extract Taxonomic IDs (domain, kingdom, phylum, class, order, family, genus, species) from taxonomic_lineage_ids column
- Keep entry, class, and sequence information
- Drop N/A
- CD-HIT using Docker at 80% similarity threshold and n=5
- Keep representative sequences
- Create labels based on classes
- Remove classes/examples with fewer than 100 samples
- Stratified split: first get test set (5,000), then eval (5,000), rest is train