Building six different drug-disease association networks. This figure shows the process of building six different drug-disease association networks. Starting with drug-specific and disease-specific subnetworks in step1, we have imported these networks into the Gephi tool to check different centrality measures for every single gene in step 2. In step 3, for each drug/disease we have selected genes with a centrality measure (degree, closeness and betweenness) that is higher than the average centrality measures for all genes. These selected genes have been used to build drug-gene and disease-gene Boolean networks. This step has been independently repeated for the three centrality measures and for the two data sources. Thus we ended up having six different drug-gene and disease-gene networks that have been used to train the regression model in step 4.