ResiRole

ResiRole assesses residue-level functional site predictions to evaluate and rank the quality of protein structure and homology models for structural biology applications.

Key Features:

• Functional Site Prediction Assessment: ResiRole evaluates protein homology models by comparing predicted functional sites at individual residues to those in experimentally determined reference structures.
• Difference Score Metric: It computes a "difference score" as the absolute value of the cumulative probability difference between FEATURE program functional site predictions on the reference structure and the model.
• Benchmarking and Validation: The method has been benchmarked against approaches including NaïveBLAST and demonstrates detection of additional local structural similarities and correlation with established model quality metrics.

Scientific Applications:

• Model selection and ranking: ResiRole ranks structure prediction techniques and models based on residue-level functional site agreement with reference structures.
• Structure-based drug design: It supports selection of models with accurate functional-site geometry for drug discovery and ligand design.
• Enzyme mechanism analysis: It aids evaluation of active-site and catalytic residue placements in predicted models.
• Protein–protein interaction studies: It assists assessment of interface residues relevant to protein–protein interactions.
• Method development and benchmarking: It provides empirical comparisons to refine and benchmark structure prediction techniques.

Methodology:

Functional site predictions are generated using the FEATURE program for both reference structures and models; per-residue cumulative probability differences are computed and their absolute values reported as difference scores, which are used to compare and rank prediction techniques and validated against benchmarks including NaïveBLAST and other model quality metrics.