GPU-based biclustering algorithm

Abstract

In this Final Year Project, it aimed to implement Geometric Biclustering (GBC) Algorithm in Graphics Processing Unit (GPU). GBC Algorithm is used to find any Maximal Biclusters, which are common patterns in each column in the gene microarray data, indicate some specific disease. A gene microarray commonly involves a huge number of data, such as thousands of rows by thousands of columns so that finding the Maximal Biclusters involves intensive computation. GPU was used mainly in graphics computing, but now it can handle the general purpose computation as CPU does. General purpose means that the results are for analysis instead of showing on the screen. The advantage of GPU is its ability of parallel computing which means that for those independent procedures, they can be carried out at the same time. Implementing this algorithm in GPU would reduce the processing time largely due to the parallel computing of GPU. Since GBC Algorithm involves large number of AND operations and GPU can carry the AND operation of the elements in each 2 different columns at the same time.