THE VARIABLE NEIGHBORHOOD SEARCH METAHEURISTIC FOR FUZZY CLUSERING cDNA MICROARRAY GENE EXPRESSION DATA

Several thousand genes can be monitored simultaneously using cDNA microarray technology. To exploit the huge amount of information contained in gene expression data, adaptation of existing and development of new computational methods are required. Recently, the Fuzzy C-Means (F-CM) method has been applied to cluster cDNA microarray data sets. To overcome some shortcomings of F-CM and to improve its performance, it was embedded into a variable neighborhood search (VNS) metaheuristic. The methodology was used to cluster four cDNA microarray data sets. Results show that VNS+F-CM substantially improves the findings obtained by F-CM. This methodology may yield significant benefit in the improvement of decision support systems used for gene expression classification.