Precise parallel volumetric comparison of molecular surfaces and electrostatic isopotentials.

Area of Science:

• Computational biology
• Structural bioinformatics
• Algorithm design

Background:

• Protein binding specificity relies on subtle geometric and electrostatic features.
• Existing algorithms lack the precision and efficiency for detailed analysis of these subtle features.

Purpose of the Study:

• Introduce pClay, the first algorithm for parallel and arbitrarily precise comparisons of molecular surfaces and electrostatic isopotentials.
• Demonstrate pClay's ability to yield more precise structural inferences than existing methods.

Main Methods:

• Developed pClay, an algorithm utilizing parallelism for high-precision geometric solid comparisons.
• Applied pClay to analyze molecular surfaces and electrostatic isopotentials.
• Evaluated pClay's performance on workstation CPUs and a 61-core Xeon Phi.

Main Results:

• pClay enables practical, arbitrarily high-precision comparisons of molecular surfaces and electrostatic isopotentials.
• Models trained with pClay data identify 100% of steric influences on specificity, compared to 47% with existing methods.
• pClay demonstrates significant parallel performance gains on multi-core processors.

Conclusions:

• pClay offers superior precision in geometric comparisons, leading to more accurate structural inferences.
• The algorithm enhances the training data for statistical models of protein binding.
• pClay has potential applications in explaining binding mechanisms and designing protein binding preferences.