The high cost and the long time required to bring drugs into commerce is driving efforts to repurpose FDA approved drugs—to find new uses for which they weren’t intended, and to thereby reduce the overall cost of commercialization, and shorten the lag between drug discovery and availability. In comparison to traditional drug repositioning, which relies on serendipitous clinical discoveries, computational methods can systemize the drug search and facilitate the drug development timeline even further. In this dissertation, I report on the development, testing and application of a promising new approach to drug repositioning. This novel computational drug repositioning method is based on mining a human functional linkage network for inversely correlated modules of drug and disease gene targets. Functional linkage network is an evidence-weighted network that provides a quantitative measure of the degree of functional association among any set of human genes. The method takes account of multiple information sources, including gene mutation, gene expression, and functional connectivity and proximity of within module genes. The method was used to identify candidates for treating breast and prostate cancer. We found that (i) the recall rate for FDA approved drugs for breast and (prostate) cancer is 20/20 (10/11), while the rates for drugs in clinical trials were 131/154 and (82/106); (ii) the Area Under the ROC Curve performance substantially exceeds that of two comparable previously published methods; (iii) preliminary in vitro studies indicate that 5/5 identified breast cancer candidates have therapeutic indices superior to that of Doxorubicin in Luminal-A (MCF7) and Triple-Negative (SUM149) breast cancer cell lines. I briefly discuss the biological plausibility of the candidates at a molecular level in the context of the biological processes that they mediate. In conclusion, our method provides a unique way of prioritizing disease causal genes and identifying drug candidates for repositioning, based on innovative computational method. The method appears to offer promise for the identification of multi-targeted drug candidates that can correct aberrant cellular functions. In particular the computational performance exceeded that of existing computational methods. The approach has the potential to provide a more efficient drug discovery pipeline.

Identifying new potential treatment options for medical conditions that cause human disease burden is a central task of biomedical research. Since all candidate drugs cannot be tested with animal and clinical trials, in vitro approaches are first attempted to identify promising candidates. Likewise, identifying other essential relations (e.g., causation, prevention) between biomedical entities is also critical to understand biomedical processes. Hence, it is crucial to develop automated relation prediction systems that can yield plausible biomedical relations to expedite the discovery process. In this dissertation, we demonstrate three approaches to predict treatment relations between biomedical entities for the drug repositioning task using existing biomedical knowledge bases. Our approaches can be broadly labeled as link prediction or knowledge base completion in computer science literature. Specifically, first we investigate the predictive power of graph paths connecting entities in the publicly available biomedical knowledge base, SemMedDB (the entities and relations constitute a large knowledge graph as a whole). To that end, we build logistic regression models utilizing semantic graph pattern features extracted from the SemMedDB to predict treatment and causative relations in Unified Medical Language System (UMLS) Metathesaurus. Second, we study matrix and tensor factorization algorithms for predicting drug repositioning pairs in repoDB, a general purpose gold standard database of approved and failed drug–disease indications. The idea here is to predict repoDB pairs by approximating the given input matrix/tensor structure where the value of a cell represents the existence of a relation coming from SemMedDB and UMLS knowledge bases. The essential goal is to predict the test pairs that have a blank cell in the input matrix/tensor based on the shared biomedical context among existing non-blank cells. Our final approach involves graph convolutional neural networks where entities and relation types are embedded in a vector space involving neighborhood information. Basically, we minimize an objective function to guide our model to concept/relation embeddings such that distance scores for positive relation pairs are lower than those for the negative ones. Overall, our results demonstrate that recent link prediction methods applied to automatically curated, and hence imprecise, knowledge bases can nevertheless result in high accuracy drug candidate prediction with appropriate configuration of both the methods and datasets used.