Statistical learning approaches to study virus evolution
Juan Angel Patino-Galindo, PhD.
Viral infections are among the most important public health concerns to humans due to their high prevalence and associated mortality. Prevention and treatment campaigns against them usually have limited efficacy. This is partly explained because their biological features allow them to reach very high levels of genetic diversity, both at the within- and between-host levels.
In this presentation I describe different, novel statistical models designed to study two relevant evolutionary phenomena in viruses. Firstly, models based on Persistent homology were applied to tens of viral species aiming to measure the occurrence of recombination. Secondly, a model that considers different levels of information from phylogenetic structure was used to predict mutations driving adaptation in HIV (resistance to antivirals) and Avian Influenza Virus (emergence to humans).