BICLUSTERING GENE EXPRESSION DATASET USING ENHANCED BARYCENTER CROSSING MINIMIZATION

There are two main categories of bi-clustering approaches graph-based bi-clustering, and non-graph based bi-clustering. In graph-based bi-clustering algorithm the input dataset is converted into bigraph such as bipartite graph, and apply some heuristic local searching techniques to minimize the number of crossings between edges in the Bigraph such as BaryCenter (BC) used in SPHier algorithm. The main problem of graph-based bi-clustering algorithm is to extract the best bi-clusters, and this leads to the ordering of gene expression dataset before we apply biclustering algorithm. In bipartite graph this is achieved through minimizing the number of crossings in bipartite graph. As we minimize the number of crossings in the bipartite graph, the gene expression dataset becomes more ordered, and this enhances the results of biclustering algorithm. The main goal of our proposed algorithm is the enhancement of graph-based biclustering algorithm by enhancing BaryCenter crossing minimization heuristics of bipartite graph. In the proposed algorithm we add the rank of each node to the rank of its neighbors, and using the position of each node in the calculations to give a new rank to each node, and using this rank for reordering the nodes of each layer.