Protein Function Training Data with GO Annotations and Interactions

Training corpus for GO-GPT, an autoregressive transformer model for Gene Ontology term prediction. It contains proteins annotated with GO terms, InterPro domains, STRING protein-protein interactions, and metadata sourced from UniProt.

Use Cases

• Train an autoregressive transformer model to predict Gene Ontology terms from protein sequences and metadata.
• Analyze the co-occurrence of InterPro domains with specific Gene Ontology term annotations.
• Build a protein-protein interaction network model using STRING data to infer functional relationships.
• Fine-tune a multimodal model that integrates protein metadata from UniProt with structured GO annotations.

Strengths

• Integrates multiple biological data modalities including GO terms, InterPro domains, and STRING interactions.
• Sourced from authoritative biological databases such as UniProt.
• Designed specifically for training a state-of-the-art transformer model for protein function prediction.

Limitations

• The specific number of protein records, features, and data volume is unknown.
• Potential class imbalance in Gene Ontology term annotations is not quantified.
• Data freshness is unclear beyond the last platform update date.

Provenance

Source

UniProt, InterPro, STRING database

Collection Method

Compiled for training the GO-GPT autoregressive transformer model.

Freshness

Last updated on the platform on 2026-03-20.