LiteFold/GOA

Gene Ontology Annotation UniProt

GOA provides high-quality, evidence-coded Gene Ontology annotations for UniProtKB proteins, RNAs, and protein complexes.

This dataset contains the original GOA UniProt source files plus a viewer-friendly Parquet sample/index table. The source goa_uniprot_all.gaf.gz and goa_uniprot_all.gpa.gz files are very large, so the default Dataset Viewer table contains the first 50,000 parsed annotation rows from each source file, along with a source-file manifest in metadata/source_files.parquet. Use the original compressed files for complete GOA coverage. Use the default Parquet table for quick inspection, schema discovery, examples, and Dataset Viewer previews.

Splits

The split is deterministic by annotation_id: sha256(annotation_id) % 10. Bucket 0 is test; buckets 1 through 9 are train.

Split	Rows
train	89,958
test	10,042
total	100,000

Source Files

File	Size
goa_uniprot_all.gaf.gz	15,387,303,487 bytes
goa_uniprot_all.gpa.gz	9,462,421,263 bytes

Usage

pip install datasets

Load the viewer table:

from datasets import load_dataset

ds = load_dataset("LiteFold/GOA")
print(ds)
print(ds["train"][0])

Load one split:

from datasets import load_dataset

train = load_dataset("LiteFold/GOA", split="train")
test = load_dataset("LiteFold/GOA", split="test")

Stream rows:

from datasets import load_dataset

stream = load_dataset("LiteFold/GOA", split="train", streaming=True)
for row in stream.take(5):
    print(row["db_object_id"], row["go_id"], row["evidence_code"])

Filter the sample for molecular-function GAF rows:

from datasets import load_dataset

ds = load_dataset("LiteFold/GOA", split="train")
mf = ds.filter(lambda row: row["source_format"] == "GAF" and row["aspect"] == "F")
print(mf[0])

Download the source manifest:

import pandas as pd
from huggingface_hub import hf_hub_download

path = hf_hub_download(
    repo_id="LiteFold/GOA",
    repo_type="dataset",
    filename="metadata/source_files.parquet",
)
source_files = pd.read_parquet(path)
print(source_files)

Download the full raw files when needed:

from huggingface_hub import hf_hub_download

gaf_path = hf_hub_download(
    repo_id="LiteFold/GOA",
    repo_type="dataset",
    filename="goa_uniprot_all.gaf.gz",
)

Columns

Column	Description
annotation_id	Stable SHA-256 ID for the sampled annotation row.
source_file	Source file: GAF or GPA.
source_format	Parsed source format, GAF or GPA.
source_row_number	Row number within the source annotation stream.
db	Source database.
db_object_id	Annotated object identifier.
db_object_symbol	GAF object symbol, when available.
qualifier	Raw qualifier field.
qualifiers	Qualifier field split on `
go_id	GO identifier.
db_references	References split on `
evidence_code	GO or ECO evidence code.
with_from	With/from field split on `
aspect	GAF aspect: F, P, or C; missing for GPA rows.
db_object_name	GAF object name, when available.
db_object_synonyms	GAF synonyms split on `
db_object_type	GAF object type, when available.
taxon_ids	GAF taxon IDs split on `
interacting_taxon_id	GPA interacting taxon ID, when available.
date	Annotation date.
assigned_by	Annotation provider.
annotation_extension	Annotation extension field.
gene_product_form_id	Gene product form identifier.
split_bucket	Deterministic split bucket from sha256(annotation_id) % 10.

Citaton

@article{huntley2015goa,
  title   = {The {GOA} database: Gene Ontology annotation updates for 2015},
  author  = {Huntley, Rachael P. and Sawford, Tony and Mutowo-Meullenet, Prudence and Shypitsyna, Aleksandra and Bonilla, Carlos and Martin, Maria Jesus and O'Donovan, Claire},
  journal = {Nucleic Acids Research},
  volume  = {43},
  number  = {D1},
  pages   = {D1057--D1063},
  year    = {2015},
  doi     = {10.1093/nar/gku1113}
}