Quadratic unconstrained binary optimization and constraint programming approaches for lattice-based cyclic peptide...

Area of Science:

• Structural Biology
• Computational Biology
• Drug Design

Background:

• Peptide-protein interactions are crucial for biological processes and drug development.
• Accurate modeling of these interactions is essential for rational drug design.
• The peptide-protein docking problem presents significant computational challenges.

Purpose of the Study:

• To explore the application of quantum-amenable Quadratic Unconstrained Binary Optimization (QUBO) for peptide-protein docking.
• To incorporate peptide cyclization and docking constraints within a QUBO framework.
• To compare the performance of a novel QUBO approach against a constraint programming (CP) method.

Main Methods:

• Developed a resource-efficient QUBO encoding for peptide-protein docking on a tetrahedral lattice.
• Implemented an end-to-end framework for evaluating QUBO and CP methods using Protein Data Bank (PDB) instances.
• Utilized a classical simulated annealing solver for the QUBO approach.
• Developed a novel constraint programming (CP) approach for benchmarking.

Main Results:

• The QUBO approach successfully modeled feasible conformations for small peptide-protein docking problems (up to 6 peptide residues).
• QUBO performance degraded significantly with increasing problem size (e.g., PDB 3WNE, 5LSO).
• The CP approach demonstrated superior scalability, solving larger instances (up to 11 peptide residues, PDB 2F58).

Conclusions:

• QUBO can be applied to peptide-protein docking but exhibits scaling limitations.
• Constraint programming shows stronger performance and scalability for larger peptide-protein docking instances.
• CP may be a more suitable method than QUBO for complex peptide-protein docking problems in drug design.