FINER

FINER predicts and refines isoform functions and isoform-isoform interactions using a unified deep learning framework to improve isoform-level functional annotation and protein-protein interaction (PPI) accuracy.

Key Features:

• Joint Prediction Framework: Combines prediction of isoform functions with refinement of PPIs from gene to isoform level, enabling mutual enhancement by leveraging shared information between the two tasks.
• Improved Predictive Performance: Demonstrates at least a 5.16% increase in Area Under the Curve (AUC) and a 15.1% rise in Average Precision Recall (AUPRC) for functional prediction across multiple human tissues by refining noisy PPIs.
• Biological Consistency: Produces predictions consistent with tissue specificity and subcellular localization of isoforms based on in-depth analyses.

Scientific Applications:

• Functional Genomics: Enhances annotation of gene products with precise isoform-level functional information.
• Proteomics: Refines PPI data to represent isoform-specific interactions for improved interaction studies.
• Disease Research: Facilitates identification of isoform-specific roles in disease mechanisms to support targeted investigation of therapeutic strategies.

Methodology:

FINER employs a unified deep learning framework that jointly predicts isoform functions and refines isoform-isoform interactions by leveraging shared data to refine noisy PPIs and improve isoform-level functional annotations.