Classification of containment hierarchy for point clouds in periodic space

When working with large point clouds, it is often useful to label the data. One of such labels is the classification of insides and outsides. Meaning that as input an a selection can be given and the output will be all that lies inside that selection. Containment labeling has shown to be very useful for point clouds in Euclidean space and allows for the generation of signed distance fields, however, such classification was not robustly available in periodic space.

Here were present our open source tool coined MDVContainment, which rigorously solves the containment problem for the periodic case. This algorithm was applied to a coarse grained acyl chains bicelle, transfecting lipoplex and a system of stacked bilayers as well as to an all atomistic periodic nanotube. Showing that the analysis is performing well—both atomistic and coarse grained systems. The containment processing of these systems takes roughly the same amount of time as creating their universe objects in MDAnalysis (i.e. reading in the data).

Having a rigorous definition of containment makes it possible for MD analysis/visualization tools to support periodic containment, just as it is supported for non-periodic spaces.