Pushing the accuracy limit of shape complementarity for protein-protein docking

Area of Science:

• Computational biology
• Structural bioinformatics
• Biophysics

Background:

• Protein-protein docking is crucial for studying molecular interactions.
• Shape complementarity is a fundamental aspect of scoring functions in docking.
• Developing accurate shape representations, especially for grid-based methods, remains a challenge.

Purpose of the Study:

• To introduce a novel pairwise shape-based scoring function (LSC) for protein-protein docking.
• To enhance the accuracy of protein-protein docking by incorporating long-range interactions.
• To improve the performance of grid-based docking algorithms.

Main Methods:

• Developed a new exponential-form scoring function (LSC) considering pairwise atomic interactions.
• Integrated the LSC scoring function into an FFT-based protein-protein docking program.
• Evaluated the LSC method on the protein docking benchmark 4.0 for both bound and unbound cases.

Main Results:

• The LSC scoring function significantly outperformed ZDOCK 2.1, MolFit/G, GRAMM, and FTDock/G in success rate and hit count.
• LSC achieved a 51.71% success rate for bound docking and 6.82% for unbound docking in top 10 predictions.
• LSC provided more hits per complex in the top 1000 predictions for both bound (8.38) and unbound (3.94) docking compared to ZDOCK 2.1.

Conclusions:

• The LSC method offers an improved initial-stage docking approach for subsequent analyses.
• The LSC scoring function can be generally implemented for accurate representation of other grid-based energy terms.
• The developed software is available via the HDOCK web server.