Molecular Surface Estimation by Geometric Coupled Distance Functions

Yuanhao Gong,Yong Chen

doi:10.1109/access.2020.3026757

Abstract

Estimating the surface from given atoms with location and size information is a fundamental task in many fields, such as molecular dynamics and protein analysis. In this paper, we present a novel method for such surface estimation. Our method is based on level set representations, which can efficiently handle complex geometries. The proposed method is analyzed from mathematical point of view and from computation point of view. The method does not require any prior information about the surface. This property is fundamentally important for the surface estimation task. The presented method is evaluated on both synthetic and real data. Several numerical experiments confirm that our method is effective and computationally efficient. Finally, the method is applied on protein surface estimation. This method is suitable for high performance molecular dynamics study, protein surface analysis, etc.

Highlights

IntroductionGiven the location and size information of some atoms, it is important to know the geometric surface that compactly contain these atoms
Given the location and size information of some atoms, it is important to know the geometric surface that compactly contain these atoms. Such geometric surface is important for its function, such as protein docking and Molecular Hydrophobicity Potential
Even though we focus on two- and three-dimensional problems, the method presented here works in higher dimensions, for example for constructing surfaces in manifold learning

Summary

Introduction

Given the location and size information of some atoms, it is important to know the geometric surface that compactly contain these atoms. Such geometric surface is important for its function, such as protein docking and Molecular Hydrophobicity Potential. Different from the molecular surface, these points exactly live on the surface and do not have a size or radius. We show this method and its accuracy on both synthetic and real data.

Objectives

Methods

Conclusion