Recognizing patterns in high-dimensional data: automated histogram filtering for protein structure elucidation

Advanced automated clustering algorithms are essential to meet the challenge of understanding high-dimensional data that arise in protein structure elucidation. In this paper, we first describe a powerful new approach to cluster analysis, automated histogram filtering (AHF). We follow this by a review of applications of AHF-clustering of Monte Carlo and molecular dynamics simulation data for off-lattice and all-atom model protein systems. Next, we present the results of a time-complexity analysis performed on a computer cluster suggesting how the computational effort scales with the size of the protein and the number of processors. And finally, we conclude with a summary of our findings.