Parallel, stochastic measurement of molecular surface area

Area of Science:

• Biochemistry and computational chemistry
• Algorithm development for scientific computing
• High-performance computing in structural biology

Background:

• Traditional molecular surface area algorithms are not optimized for modern parallel computer architectures.
• There is a need for efficient computational methods to determine protein surface areas.

Purpose of the Study:

• To develop a parallel, stochastic algorithm for molecular surface area computation suitable for multi-core architectures.
• To create a progressive algorithm that provides immediate, refining estimates of surface area.
• To enable point-based rendering of molecular surfaces.

Main Methods:

• Developed a parallel, stochastic algorithm for molecular surface area calculation.
• Implemented the algorithm on a Graphics Processing Unit (GPU).
• Compared the algorithm's performance against existing molecular surface computation programs.

Main Results:

• The parallel algorithm demonstrates suitability for multi-core architectures.
• The GPU implementation provides fast and accurate molecular surface area estimates.
• The algorithm successfully generates points on the molecular surface for rendering.

Conclusions:

• The novel parallel algorithm offers an efficient solution for computing molecular surface areas on contemporary hardware.
• The progressive nature and point generation capabilities enhance its utility in biochemistry and structural biology.
• The algorithm shows competitive performance compared to existing methods, offering speed and accuracy.