STAR-GO: Improving Protein Function Prediction by Learning to Hierarchically Integrate Ontology-Informed Semantic Embeddings

Key Points

• The aim is to enhance the prediction of protein functions through a method that integrates semantic embeddings informed by ontology.
• Developed a machine learning model called STAR-GO.
• Utilized ontology-informed semantic embeddings to train the model.
• Integrated hierarchical data to improve prediction accuracy.
• Demonstrated improvements in prediction accuracy for protein functions.
• Showed the model effectively learned from integrated semantic embeddings.