MolPathFinder: Interactive Multi-Dimensional Path Filtering of Molecular Dynamics Simulation Data

Abstract

Molecular Dynamics Simulations (MDS) play an important role in the field of computational biology. The simulations produce large high-dimensional, spatio-temporal data describing the motion of atoms and molecules. A central challenge in the field is the extraction and visualization of useful behavioral patterns from these simulations. Many visualization tools have been proposed to help computational biologists gain insight into MDS data. While recent developments focused on accelerating and optimising the rendering, it is still necessary to design new metaphors to better understand and filter MDS datasets. In this article, we are describing a set of tools to interactively filter and highlight dynamic and complex paths constituted by motions of molecules. In collaboration with computational biologists, we have tested our approach on large-scale, real data. Based on the user's feedback, our program helped scientists to navigate more easily through their dataset and isolate interesting patterns. Furthermore, our approach was useful to investigate both local and global behavior of molecular motions.