Clustering Analysis Based On The Ant System For Gene Expression Data

DNA microarray is a very powerful tool, it is a special and very meaningfultechnology after the genomics and postgenomics, and it opened a new era forexploring life, disease and drug development. At present, the focus of research forDNA microarray is no longer how to get gene expression data from the experiment,but how to mine useful information from a lot of data. Clustering is an effective toolon mining data, it has been widely applied to gene expression data processing, and itprovide the basis for predicting biological cells cycle, predicting gene function,finding pathogenic gene and explaining some new disease types.Ant colony algorithm is adopted in this paper for DNA microarray dataclustering analysis, and Fmeasure is used to evaluate the clustering result. The valueof Fmeasure is bigger, the clustering results is better. First of all, standard antclustering algorithm based on corpse cleaning, which is call LF algorithm also, is usedto analyze clustering issue of DNA microarray data, but the value of Fmeasure issmall, clustering result is not very good. The iteration process shows the LF algorithmwill lose evolution ability quickly. Secondly, inspired by LF algorithm based onentropy, This paper proposes a new LF algorithm which is based on averagedistanceThe result is not promising also. The iteration process shows the LF algorithmbased on average distance will be trapped into local optima easily. Finally, this papercombines the basic LF algorithm and the LF algorithm based on average distance withK-means algorithm to analyze the clustering items of gene expression data.Simulation results show the hybrid algorithms have good performance.