Efficient agglomerative hierarchical clustering for biological sequence analysis

Cluster analysis is an important data mining technique widely used for pattern recognition and information retrieval. In the literature, over a hundred clustering algorithms have been developed to target input datasets with different characteristics. Among these algorithms, the hierarchical clustering method is particularly useful for analyzing genetic datasets in evolutionary biology studies because of the inherent hierarchical relationships amongst the genetic sequences extracted from related organisms. However, this algorithm is computational expensive in terms of both execution time and particularly memory usage. This paper summarizes our experience in using parallel computing technologies with new algorithms to perform hierarchical sequence clustering in a more effective way without compromising the accuracy of the results.