Fast Method for Computation of Channels in Dynamic Proteins

Biochemists studying the protein properties use a computer analysis of existence and proportions of the tunnels (cavities), leading from a biochemically significant place inside a protein to its surface. In a computer simulation and visualization, a tunnel in a protein can be searched as a sequence of tetrahedra in the 3D triangulation, where the protein atoms positions are used as the triangulation vertices. The geometry of a protein is not static, the positions of atoms change in time and the biochemists have to explore a long sequence of molecule snapshots to find a stable tunnel. The recent method of a tunnel computation creates a triangulation of the whole protein for each snapshot. The method we propose uses topology information about a tunnel from the previous snapshot and a clustering of atoms to cut down the number of the triangulation vertices in the current snapshot, i.e. we compute only a triangulation of an atom subset for each snapshot. Our resulting tunnels are almost identical with the tunnels computed in the triangulation of the whole protein and the total computing time falls to thirty percent and less.