Datasets:

andrewdalpino
/

SwissProt-Gene-Ontology

dataset_info:
  - config_name: default
    features:
      - name: id
        dtype: string
      - name: sequence
        dtype: string
      - name: go_terms
        list:
          - name: id
            dtype: string
          - name: aspect
            dtype: string
          - name: evidence_code
            dtype: string
      - name: taxon_id
        dtype: int64
      - name: existence_code
        dtype: int64
    splits:
      - name: train
        num_bytes: 310253975
        num_examples: 552960
    download_size: 133898769
    dataset_size: 310253975

SwissProt Gene Ontology Dataset

This is a snapshot of the SwissProt dataset with gene ontology (GO) term annotations scraped from the Uniprot API. It is a raw and unfiltered dataset that can be processed to build datasets for protein function prediction.

Subsets

The SwissProt Gene Ontology dataset is available on HuggingFace Hub and can be loaded using the HuggingFace Datasets library.

To load the default SwissProt Gene Ontology dataset with all function annotations you can use the example below.

from datasets import load_dataset

dataset = load_dataset("andrewdalpino/SwissProt-Gene-Ontology")

References

The UniProt Consortium, UniProt: the Universal Protein Knowledgebase in 2025, Nucleic Acids Research, 2025, 53, D609–D617.