Making the most effective use of available computational methods for drug repositioning

Over the last decades, there has been substantial debate around the apparent drop in productivity in the pharmaceutical sector. The development of second or further medical uses for known drugs is a possible answer to expedite the development of new therapeutic solutions. Computational methods are among the main strategies for exploring drug repurposing opportunities in a systematic manner.

This article reviews three general approximations to systematically discover new therapeutic uses for existing drugs: disease-, target-, and drug-centric approaches, along with some recently reported computational methods associated with them.

Computational methods are essential for organizing and analyzing the large volume of available biomedical data, which has grown exponentially in the era of big data. The clearest trend in the field involves the use of integrative approaches where different types of data are combined into multipartite networks. Every aspect of computer-guided drug repositioning has currently incorporated state-of-the-art machine learning tools to boost their pattern recognition and predictive capabilities. Remarkably, a majority of the recently reported platforms are publicly available as web apps or open-source software. The introduction of nationwide electronic health records provides invaluable real-world data to detect unknown relationships between approved drug treatments and diseases.