Applying ISODATA With PCA To Identify Pathogenic Candidate Genes Associated With Alzheimer’s Disease

Alzheimer’s disease (AD) is the most common form of dementia. Atpresent, its formation mechanism is little understood. In order to understand, DNAmicroarray expression data analysis appears to be very important，and research in thisarea has made some progress. Therefore, this paper first introduces the work of theirpredecessors: analyzed gene expression data, the feature that is a strong linearcorrelation was discovered; and then using principal component analysis, thedimensions of data are reduced, and clustering feature is found among column data; andbased on it, the one-dimensional clustering algorithm is designed to estimate theinformation (class number and centers) of original data. However, we find that althoughthe first principal component has accounted for most of the information (about90%),but the remaining10%of the information is also important. Therefore, this paperpresents a new idea that using a high dimensional algorithm which can draw on theexperience of the original clustering results re-clusters it, namely the iterativeself-organizing data analysis algorithms, expecting to get an ideal clustering mode. Thenbasing on it, using the criterion of two conjugate expression and firming breaking, the37AD candidate genes are identified. In addition, we also analyze these genes bybiological and get some in-depth information on AD: The main is involved in cellsignaling (ECE1, PLD2and PHB), inflammation (il17a), ubiquitin-labeled proteindegradation (UBC) and protein synthesis (rpl32, rps6, rpl23a and rps19). In the sametime, the paper also discusses the potential impact of gene TPT1and TUBA1A. Finally,based on the above discussion, paper presents the clear pathway of potential ADpathogenesis.