The Research And Realizing Of IGA-FCM Clustering Algorithm In Gene Expression Data Analysis

A mass of gene expression data have been generated, along with the extensive application of DNA microarray technology. How to analyze and handle these data, digging out valuable biological and medical information from them, has been becoming an important hotspot at post-genomic age. Cluster analysis which groups genes with related functions according to similarities in their expression profiles, helpful to comprehend genetic function, gene regulation and cellular processes, has been a major exploratory technique.Fuzzy C-means (FCM) and Genetic algorithm (GA) are emphasized on the basis of discussing the kinds of clustering algorithms. IGA-FCM algorithm is put forward by combining improved genetic algorithm (IGA) with Fuzzy C-means (FCM), aiming at the disadvantages of FCM, sensitive to the initial value and trap easily into local high-point. The sensitivity to initial value, ability to search widely and accuracy of IGA-FCM algorithm are validated by experimenting on Wine data set, IRIS data set and Image segmentation data set respectively, and simulating in matlab. The experimental results show that the proposed algorithm has not only to some extent overcome the limitation of FCM algorithm, but also boost the speed of converging very highly and the accuracy, compared with FCM.De-noising is necessary before clustering analysis because of a lot of noise was introduced in the experimental process. Based on wavelet de-noising, IGA-FCM algorithm is applied to Yeast Cell Cycle data set. The results indicate that IGA-FCM algorithm improves the accuracy of clustering on the basis of wavelet de-noising and shorten the time of clustering. Therefore the algorithm is effective.