Enabling faster drug discovery through machine learning

In this PhD project, we will extend state-of-the-art artificial intelligence algorithms to help discover new drug molecules. The springboard for the research is a unique collaboration between Prof Jonathan Hirst (Chemistry), Prof Thomas Gaertner (Computer Science) at the University of Nottingham and the pharmaceutical company GlaxoSmithKline (GSK), which is synthesizing and testing new molecules for effectiveness as a potential treatment for a complex and chronic lung disease, idiopathic pulmonary fibrosis. We will develop new AI methods, focused on the specific demands of drug discovery. We will assess approaches such as long short-term memory, variational autoencoders, convolutional neural networks and Bayesian kernel methods. Starting from a firm theoretical foundation, we expect to derive algorithms that are more successful and more generally applicable than earlier work. Furthermore, we will evaluate the models performance by selecting a subset of predicted structures, synthesising them and testing them in established assays at GSK.