Multi-omics Data Fusion Association Analysis Based On Network Structure

The rapid development of modern society brings great mental stress to people at the same time.With the advancement of medicine,psychosis have gradually been valued by people.Schizophrenia,which is a hereditary psychotic disease among them,has attracted the attention of researchers because of its unknown etiology and complex clinical manifestations.With the great popularity of related technologies such as machine learning and gene sequencing,many researchers focus on the methods of machine learning and the omics data in biomedical science to get the information which they need.In this paper,the algorithm of multi-omics data fusion association analysis based on network structure is used to calculate and analyze the three types of omics datas of fMRI,SNP and DNA-Methy.Firstly,after preprocessing the data,we construct a network structure model to model the structure of the sample data,establish a corresponding similarity matrix for each type of data,then use the cross-diffusion process as the core algorithm which can strengthen the strong correlation and weaken the weak correlation of sample pairs.Merging each type of data into the final unified data matrix,selecting strong correlation sample pairs from the unified matrix elements,then analyzing the specific correlations.Finally,according to the corresponding SNP locus,fRMI voxel information,etc.,the biomarkers of potential schizophrenia and related information such as the potential brain area are found.Compared with other methods using linear fusion,this paper adopts the method of nonlinear information fusion,which requires less priori information and does not need to assign weights for each type of data.Through the verification of the simulated dataset and the real dataset,under the same parameters,most of the fusion sample similarity of the three types of omics data has higher similarity than the fusion samples corresponding to the two types of omics data.When looking for the potential pathogenic biomarkers of related diseases,the final analysis results of the three types of omics data are also more than the results of the analysis of the two types of omics data.On the other side it proved that the added third type of omics data have some supplements and perfection for the data fusion.To provide a certain help in the medical analysis of related diseases.