Learning and predicting the pathways of AMR evolution with hypercubic inference

Understanding and predicting how bacteria acquire antimicrobial resistance (AMR) requires us to learn about their evolutionary dynamics – a challenging problem. Existing methods for learning evolutionary pathways are limited and have yet to be widely applied to AMR. To learn how pathogens evolve AMR, we urgently need new inference approaches that can learn real evolutionary processes from real datasets, so that we can actually use the wealth of emerging clinical data to make actionable predictions.

In Bergen, we have developed HyperTraPS (hypercubic transition path sampling), an internationally acclaimed computational and mathematical approach to learn evolutionary pathways using large-scale biological data. This powerful approach has had several high-impact successes in scientific and public spheres. HyperEvol, bringing together international mathematicians, statisticians, microbiology specialists, and clinicians, will (i) make the required developments to apply this approach to AMR evolution; (ii) couple these powerful developments to emerging clinical pathogen genome data from Norway and Africa to learn and predict how AMR evolves in these cases; (iii) make these developments and findings publically and internationally available for global application. We are thus poised in an exciting position where highly feasible mathematical developments, and interdisciplinary connections, will provide a global tool for learning how AMR evolves in different pathogens, populations, and countries.

Given a new strain of a pathogen observed in the clinic, with a particular drug resistance profile, we will be able to predict the most likely drugs to which it will next develop resistance, and tailor clinical decisions and treatment strategies accordingly. We will be able to analyse whether geography, demography, or other features shape AMR evolution in given pathogens, or whether evolutionary pathways are “universal”. We will also learn basic biology – which genetic and/or drug resistance features increase (or decrease) the probability of others evolving. 

Meet the project leader

© Øystein Rygg Haanæs
© Øystein Rygg Haanæs

Iain Johnston

Iain is a professor in the Department of Mathematics and an associate group leader at the computational biology unit. His research focuses on using mathematical and statistical approaches to learn about biomedical processes, particularly those that involve a random aspect (like evolution). He is an ERC starter grantee, President’s Medallist of the Society of Experimental Biology, and his work has recently been featured in the New York Times, Le Monde, and many other international media.

Learn more about HyperEvol

Watch Iain's video-presentation explaining the project during CAMRIA's opening:

Do you want to join the HyperEvol-team?

Apply for a postdoc position connected to the project (application deadline: August 1, 2022):