# ProteInfer Swiss-Prot Training Sets for EC Number and Gene Ontology Prediction (Random and Clustered Splits)

## Facts

- Provider: Google Research (Sanderson et al., ProteInfer)
- License: CC-BY-4.0 (https://storage.googleapis.com/brain-genomics-public/research/proteins/proteinfer/README.md)
- Format: TSV (in sampled files)
- Stored at: Google Cloud Storage, gs://brain-genomics-public/research/proteins/proteinfer (us)
- Access: public, readable without an account
- DOI: https://doi.org/10.7554/eLife.80942
- Category: Proteomics
- Source: https://doi.org/10.7554/eLife.80942
- Page: https://scigantic.com/archive/6796d4bb-2b1b-587c-802d-0e5d0aaae99a
- Keywords: Proteomics, TSV, Google Cloud Storage

## About

Training and evaluation data for ProteInfer, the convolutional networks that predict protein function directly from amino acid sequence (Sanderson et al., ProteInfer, deep networks for protein functional inference, eLife). Sequences come from Swiss-Prot, the manually curated part of UniProt, labeled with Enzyme Commission (EC) numbers and Gene Ontology (GO) terms, with parent terms added so the labels follow each hierarchy. Two splits are given: a random 80/20 split, and a clustered split by UniRef50 cluster membership so that test sequences have low similarity to training sequences. Each TFRecord example carries the UniProt accession, the sequence and its list of database cross-reference labels. Measured 2026-11-17: datasets/swissprot/ holds random/ and clustered/, plus price_annotations.tsv; sibling folders hold models/, blast_baseline/, interpro_baseline/, swissprot_inference_results/ and colab_support/. Read the records with TensorFlow tf.data. Built from UniProt data; the folder README licenses it CC BY 4.0. Cite Sanderson et al., eLife, and UniProt.

## Load the data

```python
import gcsfs
fs = gcsfs.GCSFileSystem(token='anon')
fs.ls('brain-genomics-public/research/proteins/proteinfer/', detail=False)[:10]
```

Lists the files with anonymous access. Runs on any machine, with no cloud account.

## Data structure

File types in sampled files: TSV

Example file: `research/proteins/proteinfer/blast_baseline/blast_output/clustered/blast_out_dev.tsv`

## Open it

- In a hosted notebook with the data mounted: https://scigantic.com/archive/6796d4bb-2b1b-587c-802d-0e5d0aaae99a
- Full record as JSON (schema card included): https://api.scigantic.com/api/archives/6796d4bb-2b1b-587c-802d-0e5d0aaae99a
- Over MCP: `get_schema_card` with id `6796d4bb-2b1b-587c-802d-0e5d0aaae99a` at https://api.scigantic.com/mcp

## Related topics

- [TSV datasets](https://scigantic.com/topics/tsv)
- [Proteomics datasets](https://scigantic.com/topics/proteomics)
