Optimal Function Prediction of Key Aberrant Genes in Early-onset Preeclampsia Using a Modified Network-based Guilt by Association Method

To predict the optimal functions of key aberrant genes in early-onset preeclampsia (EOPE) by using a modified network-based gene function inference method. First, differentially expressed genes (DEGs) were extracted using linear models for microarray data (LIMMA) package. Then the Spearman's rank correlation coefficient was calculated to assess co-expressed strength of each interaction between DEGs, based on which the co-expressed genes network was constructed to vividly exhibit their interlinking relationship. Subsequently, Gene ontology (GO) annotations for EOPE were collected according to known confirmed database and DEGs. Ultimately, the multifunctionality algorithm was used to extend the “guilt by association” method based on the co-expressed network, and a 3-fold cross validation was operated to evaluate the accuracy of the algorithm. During the process, the GO terms, of which the area under the curve (AUC) over 0.7 were screened as the optimal gene functions for EOPE. Six functions including the ion binding and cellular response to stimulus were determined as the optimal gene functions. Such findings should help to better understand the pathogenesis of EOPE, so as to provide some references for clinical diagnosis and treatment in the future.