Research On Weighted Two - Way Clustering Algorithm Based On Gene Expression Microarray Datasets

With the rapid development of biological information technology, gene chip technology has the obvious progress in the field of biology, it not only reflects the current cell physiological condition and the correlation between gene, but also for cancer subtype identification, drug targets detection, the diagnosis of drug efficacy,and reveal the disease mechanisms, and other fields play a crucial role, so based on some related gene expression data analysis is the research in the field of biological science.Because the gene expression data has two-way correlation, so the traditional clustering algorithm in the analysis of gene expression data has come under a certain limit, the traditional clustering algorithm is only a single row or column for clustering gene expression matrix to find the global information, and the final ignores one of the important local information, so the general clustering algorithms is no way to accurately reflects the close relationship between genes and conditions.In this paper, aiming at the shortcomings of the general clustering method, is proposed based on gene expression data of unsupervised weighted two-way clustering algorithm, specific done the following work: first, using the minimum spanning forest law generated the gene regulation networks. And according to the importance of gene in the gene regulatory network give the corresponding weights of every gene.Secondly, based on the gene regulation network proposed the weighted clustering algorithm. Finally, for the existing validation index, namely the average residual index(ASR) and then will improve, ultimately determine the optimal clustering number. And we apply this method to breast cancer and juvenile rheumatoid arthritis on two groups of data, examples show that this paper proposed weighted two-way clustering algorithm has good clustering effect.