Building Molecular Interaction Networks from Microarray Data for Drug Target Screening.

Potential drug targets for the disease treatment can be identified from microarray studies on differential gene expression of patients and healthy participants. Here, we describe a method to use the information of differentially expressed (DE) genes obtained from microarray studies to build molecular interaction networks for identification of pivotal molecules as potential drug targets. The quality control and normalization of the microarray data are conducted with R packages simpleaffy and affy, respectively. The DE genes with adjusted P values less than 0.05 and log fold changes larger than 1 or less than -1 are identified by limma package to construct a molecular interaction network with InnateDB. The genes with significant connectivity are identified by the Cytoscape app jActiveModules. The interactions among the genes within a module are tested by psych package to determine their associations. The gene pairs with significant association and known protein structures according to the Protein Data Bank are selected as potential drug targets. As an example for drug target screening, we demonstrate how to identify potential drug targets from a molecular interaction network constructed with the DE genes of significant connectivity, using a microarray dataset of type 2 diabetes mellitus.