Finding the Optimal Gene Order in Displaying Microarray Data

The rapid advances of genome-scale sequencing have brought out the necessity of developing new data processing techniques for enormous genomic data. Microarrays, for example, can generate such a large number of gene expression data that we usually analyze them with some clustering algorithms. However, the clustering algorithms have been ineffective for visualization in that they are not concerned about the order of genes in each cluster. In this paper, a hybrid genetic algorithm for finding the optimal order of microarray data, or gene expression profiles, is proposed. We formulate our problem as a new type of traveling salesman problem and apply a hybrid genetic algorithm to the problem. To use the 2D natural crossover, we apply the Sammon’s mapping to the microarray data. Experimental results showed that our algorithm found improved gene orders for visualizing the gene expression profiles.