Despite recent advances in life sciences and technology, the amount of money spent developing a single drug has stayed drastically expensive. Overall efficiencies have caused drug development to stay the same, with an average cost of $2.6 billion and 15 years to develop a single drug. Considering these challenges, there is an increased need for drug repositioning, in which new indications are found for existing or unapproved drugs. Here we introduce an approach that integrates only drug similarity metrics, such as side effect, structure, and target similarities, to identify novel indications for drugs. By focusing on drug similarity metrics, our proposed method allows for applications towards orphan molecules that presently have no primary indication. To improve upon the current methods of drug repurposing, we propose the developing of a computational approach that utilizes multiple data types within a machine-learning framework in order to predict indications a drug may treat. Based on the observations that similar drugs are used for similar indications, this method utilizes publicly available databases to identify associations between drugs, and integrates drug similarity data, as well as drug-target specific information, into a machine-learning framework in order to accurately predict indications for these drugs. Altogether, our method provides a novel, broadly applicable strategy that can identify novel indications, allowing for an accelerated and more efficient method for future drug development and repositioning efforts.
Bringing a drug to market requires an enormous amount of time and money, so our proposed research is designed to address this unmet need to streamline the drug discovery process. Using big data analysis, we are developing a computational method to utilize drug similarity information to predict novel diseases that drugs can treat.