This is a nice paper from my old adviser that attempts to identify what host a virus can infect based on its genome using multiclass Adaboost. It'd be cool to reproduce this with work with a DNN. Also the author is currently a postdoc at Stanford.
Furthermore, we could expand on this to find more subtle patterns. For example, this was a really cool paper from UCSF that found ball pythons being infected with a virus of possible mouse-origin. If we could make a model that can predict the potential of these sorts of cross-species events (a.k.a zoonoses), it would be a game-changer for viral epidemiology, especially for diseases like influenza.
This is a nice paper from my old adviser that attempts to identify what host a virus can infect based on its genome using multiclass Adaboost. It'd be cool to reproduce this with work with a DNN. Also the author is currently a postdoc at Stanford.
Furthermore, we could expand on this to find more subtle patterns. For example, this was a really cool paper from UCSF that found ball pythons being infected with a virus of possible mouse-origin. If we could make a model that can predict the potential of these sorts of cross-species events (a.k.a zoonoses), it would be a game-changer for viral epidemiology, especially for diseases like influenza.