Unveil fitness landscapes of newly discovered zoonotic viruses,
contributing to preventive medicine
Having this experimental evolution model mentioned above in hand, we can
further use this system, especially while dealing with emerging- and
newly discovered zoonotic viruses, to in vitro generate mutations
by reiterating several times of infection in natural host cells/animals
followed by testing viral fitness landscapes of variant strains in human
cells, thereby helping us to understand and predict the possible impact
of variant strains derived from these emerging- and newly discovered
zoonotic viruses on humans. Retrospective studies of constructing
evolutionary trajectories based on phylogenetic analyses should be
performed; in such a way, we can closely monitor the sequential
transition of virus sequence changes reshaped by evolutionary
constraints throughout every passage of cells or every filial generation
in animal models at a single-sequence level and estimate the
evolutionary timescale that requires viruses to reprogram their
tropisms. This experimental evolution model is inspired by the
evolutionary approach named
retrovolution[153],
with which the authors studied the human deoxycytidine
kinase[153].
Here under the framework of this approach with modifications, we import
the idea of incorporating molecular barcodes embedded in our viruses of
interest, thereby bringing our investigation down to the resolution of a
single virus sequence. Most importantly, in line with the procedure
designed in this model we will be able to immediately assay viral
fitness enhanced by emerging mutations that can be traced through
phylogenetic trees. In other words, this barcoding evolution model
enables us to associate genotypic changes occurring in natural host
cells/animals with predictive phenotypes in human cells in every single
evolutionary event in order to accelerate our understanding of newly
discovered zoonotic viruses and benefit to develop antiviral strategies
in advance. It is worthy to stress the point that in addition to
different experimental approaches can be used for predicting the
likelihood of predictive
mutations[152,154–156],
we also propose to use in silico approaches for this purpose: we
suggest that algorithms, like generative adversarial
networks[157], an architecture for training
generative models, can be for instance applied for predicting
evolutionary paths when the number of input sequences is sufficient.
Eventually virus sequences with predictive mutations will also be
necessary to test for fitness in human cells.