Identifying Differential Gene Networks Based Bagging Dtrace Model

The pattern of gene regulatory networks under different pathological conditions has changed,indicating that the genetic link has changed.This paper attempts to find the key genes that change the connection relationship to distinguish different subtypes of cancer.From the study of historical literature,it is common to use direct and indirect methods to estimate the difference network.Considered the time-consuming and accuracy,the directed methods are always used to estimated the differential networks.Due to the lack of data,these methods cannot fully utilize the information of data.Based on this,the text proposes the use of Bagging algorithm to estimate the difference network,through the replacement of the sample,increasing the diversity of the data and making full use of the data.The model and the Dtrace model are compared on the simulation data.Simulation results show that the proposed model is better than the Dtrace model.In real-life datasets analysis,our method was applied to the analysis of gene expression data of different subtypes of breast cancer.The stable selection algorithm was used to select the parameters,and finally 5 key genes were derived from the model.Through investigation and research,we found that these 5 hub genes play an important roles in distinguish twodifferent subtypes of breast cancer.