Unravelling human-pathogen co-evolution with genomics and deep learning

Pathogens and infectious diseases have been the most powerful drivers in human evolution. However, the extent to which past epidemics shaped the diversity of immune-related genes is still under extensive debate. In fact, current methodologies do not have sufficient power to detect signals of recent pathogen-driven selection.

A solution is given by the joint analysis of genomes pre- and post-epidemics, as extensive ancient DNA data sets are now widely accessible. Additionally, tracking how the genomes of human pathogens change in time will give us notable information of co-evolution trajectories. However, the integration of these sources of information is computationally hard. Recently, deep learning algorithms, trained with simulations, have been successfully applied to population genomic data to infer complex evolutionary histories.

The project will utilise contemporary and ancient genomes of both human populations and pathogens to understand the dynamics of their co-evolution by detecting joint signals of natural selection. A novel deep learning inferential framework will be developed for this purpose. Applications will range from Plasmodium sp. in Africa and South America, to Yersinia pestis in Europe and South-Western Asia. Findings from this project will elucidate how humans and pathogens co-evolved during the past few thousand years, and provide us with better tools to predict susceptibility to future epidemics.