Domain2GO

Domain2GO predicts protein domain functions by inferring associations between protein domains and Gene Ontology (GO) terms from co-annotation patterns and propagating those domain-associated GO terms to proteins to support functional annotation.

Key Features:

• Mutual Annotation-Based Prediction: Infers associations between protein domains and GO terms by analyzing co-annotation patterns within protein-level GO annotations and domain annotations.
• Statistical Resampling for Reliable Associations: Employs statistical resampling techniques to assess the robustness of observed domain–GO co-annotation associations.
• Biological Relevance Evaluation: Selected domain–GO mappings were evaluated for biological relevance via literature review.
• Function Prediction Propagation: Propagates domain-associated GO terms to proteins annotated with those domains to predict unknown protein functions.
• Performance Evaluation and Comparison: Performance was evaluated using CAFA3 datasets, demonstrating potential in predicting molecular function and biological process terms with interpretable results and low computational cost.
• Extensibility to Other Ontologies: Methodology can be extended to other ontologies and biological entities to investigate unknown relationships in large-scale biological data.

Scientific Applications:

• Functional Annotation: Supports annotation of protein function by assigning GO terms based on domain associations.
• Disease Research: Enables generation of protein function hypotheses that can inform studies of disease mechanisms and potential targets.
• Biological Process Elucidation: Facilitates exploration of biological processes at the domain and protein level through inferred GO term associations.

Methodology:

Analyze co-annotation patterns between protein-level GO annotations and domain annotations, apply statistical resampling to test association robustness, propagate domain-associated GO terms to proteins annotated with those domains, and evaluate performance using CAFA3 datasets.