Webinar | AI and Network Medicine approaches for predicting drugs for Covid-19

Abstract: The nature of the COVID-19 pandemic requires the rapid development of new therapies. Drug repositioning is a drug discovery strategy that exploits already existing drugs and can therefore dramatically reduce the time and costs of developing a drug de novo.  However, the experimental identification of clinically approved drugs that may have a therapeutic effect in COVID-19 patients is a complex endeavour, that is made even more difficult by our limited knowledge of the virus. Therefore, computational approaches that can predict novel testable hypotheses for systematic drug repositioning are appealing as they could potentially be used to focus and prioritize wet lab experiments.

In this talk, we will present some very preliminary results of our ongoing work in the areas of network medicine and machine learning to develop new algorithms for the prediction of repositionable drugs for Covid-19. We are following two different approaches.

A first systems pharmacology-based approach, aims at predicting drugs effective against SARS-CoV-2. Our starting point is a recently released dataset, that summarizes the state of development of several safe-in-man broad spectrum antiviral agents. Our model assumes that there exist a small set of latent features that can characterise the drug-virus interplay, and we frame the problem as a matrix decomposition problem – this is similar in flavour to a method that we developed recently for the prediction of the frequency of drug side effects in the population.

A second network medicine-based approach, aims at predicting drugs that act on human proteins, rather than on viral proteins, perturbing the human protein interaction network (interactome) in areas that are crucial for SARS-CoV-2 infection (host proteins). By calculating network diffusion distances on the interactome between these host proteins and drugs targets from DrugBank, we aim to identify lists of drugs that can perturb the areas involved in the infection.

Short bio: Alberto Paccanaro completed his undergraduate studies in Computer Science at the University of Milan and received his PhD from the University of Toronto in 2002, specializing in machine learning under the supervision of Geoffrey Hinton. From 2003 to 2006 he was a postdoc in Mark Gerstein’s lab at Yale University. In 2006, he moved to the Department of Computer Science at Royal Holloway University of London where he became full Professor in Machine Learning and Computational Biology and Director of the Centre for Systems and Synthetic Biology. In February 2020, he moved to FVG EMAp where he is currently full professor. His main research interests are in applying and developing machine learning algorithms for solving problems in molecular biology, medicine and pharmacology.