DockQ: A Quality Measure for Protein-Protein Docking Models

Area of Science:

• Computational biology
• Structural bioinformatics
• Protein structure prediction

Background:

• Current protein-protein docking model assessment relies on independent measures like Fnat, LRMS, and iRMS.
• These measures, while informative, are often used with ad-hoc cutoffs, limiting detailed analysis and model ranking.
• The CAPRI criteria classify models into four broad categories, hindering fine-grained quality evaluation and correlation with scoring functions.

Purpose of the Study:

• To introduce DockQ, a novel, continuous quality measure for protein-protein docking models.
• To provide a single, quantitative score integrating Fnat, LRMS, and iRMS for improved model assessment.
• To enable more precise ranking and correlation of docking models with scoring functions.

Main Methods:

• Developed DockQ by combining Fnat, LRMS, and iRMS into a single score ranging from 0 to 1.
• Validated DockQ using existing CAPRI benchmark datasets.
• Evaluated DockQ's ability to reproduce CAPRI classifications and its potential for quantitative analysis.

Main Results:

• DockQ successfully reproduces the CAPRI classification with high accuracy (94% PPV at 90% Recall).
• The continuous nature of DockQ allows for higher resolution assessment compared to the discrete CAPRI categories.
• DockQ facilitates quantitative analysis, such as Z-scores, and direct correlation with scoring functions.

Conclusions:

• DockQ offers a superior, continuous metric for assessing protein-protein docking model quality.
• It serves as a higher-resolution alternative to the CAPRI classification system.
• DockQ is expected to aid in the development of improved scoring functions for protein docking.