Accurate identification and volume computation of molecular interaction regions over 3D triangular meshes.

Abstract

To facilitate the rapid prediction of optimal conformations of chiral catalysts and substrates, we previously proposed two slicing-based methods for computing interacted volumes between molecules. However, the accuracy of these methods is limited by slice spacing, and they inevitably lose 3D interaction information. In this paper, we propose a method that directly operates on triangular mesh models of molecular electrostatic potential isosurfaces to compute interacted volumes with higher accuracy while preserving full 3D geometry. Our approach combines three core components: a composite bounding box that efficiently prunes non-intersecting facets; a 2D constrained Delaunay remeshing strategy that resolves intersecting triangles without generating new degenerate elements; and a signed-volume summation that leverages consistent normal orientations to compute the final volume. Although each underlying technique is standard in computer graphics, their systematic combination and adaptation to molecular electrostatic potential isosurfaces is new. Experimental results show that our method effectively removes non-interacted facets across different molecular datasets, with average removal rates of 84.57%, 86.75%, and 85.61%. Under the same conditions, it achieves an average relative error of 0.0013%, substantially outperforming previous slicing-based methods (1.73% and 0.07%). This work provides a high-accuracy and fully 3D tool for evaluating spatial compatibility in catalyst-substrate pairs.