"PADRE: Prediction Antiviral Drug Resistance before its Emergence"

Antiretroviral drug resistance (ADR) mutations are currently investigated either in vitro, using expensive experiments, or in vivo, in a population of naive or treated individuals. The in vitro approach, on the one hand, does not represent a real-world scenario, as experimental conditions yield only a biased perspective, with a limited representation of in-patient virus evolution (e.g., different subtypes and individual variation regarding immune response). On the other hand, in vivo investigation offers only the possibility to track mutations after its emergence, i.e., when it is too late to benefit public health prevention measures. Until now, investigation of HIV-1 drug resistance has relied mostly on clinical research and data mining approaches, with a limited role of artificial intelligence.
Given the recent advances in artificial intelligence, there is an important potential to study drug

resistance towards antiviral compounds through a combination of computational methods that takes into account viral evolution, protein structures and molecular docking. Such an approach would facilitate the prediction of drug resistance mutations directly in silica.


Our overall objective is to develop a computational method to predict antiviral resistance mutations towards future emerging viruses or viral variants, taking into account the effect of selective pressure of antiviral compounds, by considering the virus' protein and drug's molecular structure.
The specific objectives include to predict the protein folding of viral variants, to simulate the docking of viral protein with antiviral compounds, to calculate fitness functions of viral variants in the presence of antiviral compounds, and to use this fitness function to simulate the evolution of viral variants and potential emergence of antiviral resistance mutations.
While it is our goal to develop a method that can be used to predict the emergence of novel resistance mutations in the context of different viruses, we will evaluate the method in the context of HIV. HIV has been consistently used as a research model in several contexts, given the abundance of data available for HIV research and the knowledge that has been gathered along the years.