Study Of Feature Gene Analysis Based On The Network Module

The gene expression profile is often used for identification and diagnosis of cancer.Gene chip technology can study gene expression profiles to large-scale with high flux of cancer and overcome the limitation of chain and association studies between the disease and only one or a few candidate genes.The technology has been widely applied to many fields.The data gene expression microarray has generated with high-dimension,high-noise,high redundancy features,and less samples and most has nothing to do with disease classification.So feature gene selection can not only help us find significant genes with discriminant ability,but also can reduce the computation complexity of time and space.So the genetic selection is quite important.The main research work is as follows:Common gene selection methods considered only the independence of the individual genes without considering the interaction between genes.This paper presents a weights gene selection method of gene co-expression based on the network and used for cancer identification and diagnosis.Above all,in cancer data sets dissimilar coefficients of different nodes can be computed based on the Pearson correlation coefficient between genes,and weight network can be built.Genes have the high degree of common expression within the same module while has the low degree of common expression between different modules.Then,according to the module eigenvector the correlation between module and specific phenotype can be researched and significant module associated with the disease can be found.Three public gene expression profile datasets are used to test our method.We use the decision tree and support vector machine(SVM)as the classifier to verify the prediction accuracy of classification of the candidate genes.Experiment results show that our method can achieve good classification performance.For the neural mechanisms of schizophrenia,first,respectively to the three dimension reduction after gene expression profile data sets in different regions construction of network module.Then each module to different brain regions between the two comparison analysis.Find out the differences between regions,and the differences between these modules for network visualization analysis.And then the significant genes has been found are mapped to GO analysis in the human genetic databases.Finally we can draw the conclusion that basal ganglia area is the center of schizophrenia gene expression.For the neural mechanisms of schizophrenia,first of all,respectively to the three different brain regions gene expression profile data sets to construction of network module.Then for each modules of different brain regions to comparison analysis each other.Find out the differences for each region,and the differences between these modules for network visualization analysis.And then to find out the significant genes mapped to GO analysis in the human genetic databases.Finally draw the conclusion that basal ganglia area is the center of schizophrenia gene expression.