Geometric Biclustering Analysis of DNA Microarray Data based on Hypergraph Partitioning

Biclustering can perform simultaneous pattern classification in both row and column directions in a data matrix and is useful for DNA microarray data analysis. In this paper, a new biclustering method is introduced based on a geometrical method of identifying bicluster patterns. The Hough transform in column-pair space is used to find sub-biclusters and a hypergraph model is used to merge the sub-biclusters into larger ones. The hypergraph based geometric biclustering (HGBC) algorithm proposed here reduces the computing time and improves the classification accuracy considerably compared with exiting biclustering methods. Experiments on both simulated and real microarray data demonstrate that our method can identify biclusters with different noise levels and overlapped degrees.