An Advanced Hierarchical algorithm for protein complex prediction

High throughput experimental methods which detect protein-protein interactions have generated large datasets offering a first estimation and representation of an organism's protein interaction network. However, there is still lack of information concerning protein complexes, although many automated methods have been applied to this problem. In this paper, a new hierarchical clustering algorithm, called Advanced Hierarchical Clustering (AHC) algorithm, is proposed which detects protein complexes with high predictive ratio. The main advantage of our algorithm is the accuracy of prediction of the protein complexes from the initial protein interaction graphs. We present experimental results using 7 experimental datasets and compare them with those derived from other existing algorithms (such as Mcode, HCS, RNSC and SideS), to demonstrate the efficiency of the AHC regarding successful prediction ratio of protein complexes and accuracy.