A novel entropy-based hierarchical clustering framework for ultrafast protein structure search and alignment

Identification and alignment of three-dimensional folding of proteins may yield useful information about relationships too remote to be detected by conventional methods, such as sequence comparison, and may potentially lead to prediction of patterns and motifs in mutual structural fragments. With the exponential increase of structural proteomics data, the methods that scale with the rate of increase of data lose efficiency. Hence, new methods that reduce the computational expense of this problem should be developed. We present a novel framework through which we are able to find and align protein structure neighbors via hierarchical clustering and entropy-based query search, and present a web-based protein database search and alignment tool to demonstrate the applicability of our approach. The resulting method replicates the results of the current gold standard with a minimal loss in sensitivity in a significantly shorter amount of time, while ameliorating the existing web workspace of protein structure comparison with a customized and dynamic web-based environment. Our tool serves as both a functional industrial means of protein structure comparison and a valid demonstration of heuristics in proteomics.