Applying PCA And Hierarchical Clustering To Identify Candidate Genes Associated With Alzheimer’s Diseases

Alzheimer’s disease (AD) is one of the most common forms of dementia.The disease is degenerative and can’t be cured. It is estimated that Alzheimer’s willaffect1in85people globally in forty years. Several hypotheses are proposed to try toexplain the cause of AD, although people know a little about Alzheimer’s disease. Thewidespread hypothesis is the amyloid hypothesis. Now, people are success inidentifying three early-onset familial AD genes (APP, PSEN1and PSEN2) and onelate-onset genetic risk factor of AD (APOE). Identifying more AD candidate genesassociated with AD becomes an important research. At present, it becomes a researchfocus that identifying AD candidate genes by computation based on DNA microarraydata. Principal component analysis (PCA) is a more effective method to identify datapattern. And cluster analysis is also a scientific and effective method to study thesimilarities of objects. In this paper, we will apply these two methods to identify ADcandidate genes.Using PCA to process gene expression levels, some features are observed. Andaccording to these features, co-regulated genes are clustered together by splithierarchical clustering.168co-regulated genes of above four AD genes and68isolatedpoints are identified. What’s more,4co-regulated genes have been confirmed toassociate with AD by other references. On the other hand, the clustering effect ofhierarchical cluster in this paper is much better than hierarchical cluster in ClusterSoftware. The better cluster effect and four proved co-regulated genes both suggest thatalgorithm in this paper is more effective.Analyzing the identified genes, some possible causes of AD are proposed. Thesecauses postulate that Alzheimer’s disease is a process of apoptosis because of theweakening activities of proteins, which are mainly located at membrane and nucleus of neurons and the activities of which are mainly correlative with signal transduction andregulation of transcription possibly. Finally, potential implications of seven ADcandidate genes are analyzed.