AlphaFoldDB / dataset_summary.json
anindya64's picture
Add normalized Parquet train/test accession index
1f9dba5 verified
{
"source": "LiteFold/AlphaFoldDB",
"table_source_file": "accession_ids.csv",
"split_strategy": "deterministic hash(uniprot_accession) % 10; bucket 0 is test, buckets 1-9 are train",
"splits": {
"train": 222017452,
"test": 24672064
},
"total_rows": 246689516,
"sequence_length": {
"min": 5,
"median_approx": 278,
"mean": 328.55367611974236,
"max": 4186
},
"fragmented_predictions": 0,
"rows_without_fragment_number": 5619027,
"latest_version": {
"min": 1,
"max": 6,
"counts": {
"1": 5271725,
"2": 347302,
"6": 241070489
}
},
"parquet_files": {
"train": 12,
"test": 2
},
"columns": [
"uniprot_accession",
"alphafold_id",
"latest_version",
"first_residue_index",
"last_residue_index",
"sequence_length",
"fragment_number",
"is_fragmented_prediction",
"split_bucket"
],
"archive_metadata": {
"archive_rows": 48,
"archive_types": {
"global_health": 30,
"proteome": 16,
"swissprot": 2
},
"archive_predicted_structures": 1656090,
"archive_size_bytes": 153231764480
}
}