Quantitative analysis of molecular surface based on improved Marching Tetrahedra algorithm

Abstract

Quantitative analysis of molecular surface is a valuable technique for analyzing non-covalent interaction, studying molecular recognition mode, predicting reactive site and reactivity. An efficient way to realize the analysis was first proposed by Bulat et al. (J. Mol. Model., 16, 1679), in which Marching Tetrahedra (MT) approach commonly used in computer graphics is employed to generate vertices on molecular surface. However, it has been found that the computations of the electrostatic potential in the MT vertices are very expensive and some artificial surface extremes will be presented due to the uneven distribution of MT vertices. In this article, we propose a simple and reliable method to eliminate these unreasonably distributed surface vertices generated in the original MT. This treatment can save more than 60% of total analysis time of electrostatic potential, yet the loss in accuracy is almost negligible. The artificial surface extremes are also largely avoided as a byproduct of this algorithm. In addition, the bisection iteration procedure has been exploited to improve accuracy of linear interpolation in MT. The most appropriate grid spacing for surface analysis has also been investigated. 0.25 and 0.20bohr are recommended to be used for surface analysis of electrostatic potential and average local ionization energy, respectively.