Date of Graduation


Document Type


Degree Type



Statler College of Engineering and Mineral Resources


Lane Department of Computer Science and Electrical Engineering

Committee Chair

Donald Adjeroh

Committee Co-Chair

Donald Adjeroh

Committee Member

Donald Adjeroh

Committee Member

Gianfranco Doretto

Committee Member

Wanhong Zheng


Adverse drug events represent a key challenge in public health, especially with respect to drug safety profiling and drug surveillance. Drug-drug interactions represent one of the most popular types of adverse drug events. Most computational approaches to this problem have used different types of drug-related information utilizing different types of machine learning algorithms to predict potential interactions between drugs. In this work, our focus is on the use of genetic information about the drugs, in particular, the protein sequence and protein structure of drug protein targets to predict potential interactions between drugs. We collected information on drug-drug interactions (DDIs) from the DrugBank database and divided them into multiple datasets based on the type of information, such as, chemical structure, protein targets, side effects, pathways, protein-protein interactions, protein structure, information about indications. We proposed a similarity-based Neural Network framework called protein sequence-structure similarity network (S3N), and used this to predict the novel DDI’s. The drug-drug similarities are computed using different categories of drug information based on multiple similarity metrics. We compare the results with those from the state-of-the art methods on this problem. Our results show that proposed method is quite competitive, at times outperforming the state-of-the-art. Our performance evaluations on different datasets showed the predictive performance as follows: Precision 91\%-98\%, Recall 90\%-96\%, F1 Score 86\%-95\%, AUC 88\%-99\% Accuracy 86\%-95\%. To further investigate the reliability of the proposed method, we utilize 158 drugs related to cardiovascular disease to evaluate the performance of our model and find out the new interactions among the drugs. Our model showed 90\% accuracy of detecting the existing drug interactions and identified 60 new DDI’s for the cardiovascular drugs. Our evaluation demonstrates the effectiveness of S3N in predicting DDI’s.