3D-Garden: a system for modelling protein-protein complexes based on conformational refinement of ensembles generated...

Area of Science:

• Computational Biology
• Structural Biology
• Bioinformatics

Background:

• Accurate protein-protein complex modeling is crucial for drug design and understanding biological functions.
• Existing methods often lack backbone flexibility and integration of experimental/bioinformatic data.
• The 3D-Garden system addresses these limitations for enhanced protein docking.

Purpose of the Study:

• To develop and present 3D-Garden, a flexible protein-protein docking system.
• To incorporate backbone flexibility and user-defined restraints into the docking process.
• To provide public access to the 3D-Garden server for broader scientific use.

Main Methods:

• 3D-Garden utilizes a facet representation of molecular surfaces for user-defined patch programming.
• Flexibility is implemented via a weighted exhaustive conformer search for clashing molecular branches.
• The system filters an initial pool of ~340,000 models down to 5000 for analysis.

Main Results:

• On Benchmark 2.0, 3D-Garden achieved acceptable or better models in 29/45 cases, including one with significant conformational change.
• In 19/45 cases, the top-ranked model was acceptable or better.
• The system demonstrates robust performance in identifying accurate protein-protein complex structures.

Conclusions:

• 3D-Garden provides a robust and flexible approach to protein-protein docking.
• The system's ability to handle conformational flexibility enhances modeling accuracy.
• Public availability of the 3D-Garden server facilitates research in structural biology and drug discovery.