By Greg Mercer
I attended ASM BioThreats 2017’s panel “Predicting Emergence by Understanding the Past: Methods that Move Us towards Predictive Biology,” where a panel of researchers presented their recent efforts to get ahead of the evolutionary curve and anticipate new developments in infectious disease.
Marco Vignuzzi, of the Pasteur Institute, described his efforts to monitor, predict, and target RNA virus evolution. RNA viruses mutate constantly; any response to them must into account incremental changes and variations. Vignuzzi described a large population of many low-frequency mutants as a quasi-species or “cloud.” One can sequence the average genetic profile of this cloud, known as the “consensus sequence.” This population exists across a fitness landscape, ranging from well-adapted to poorly-adapted. The natural evolutionary tendency of a fast-mutating RNA virus is to “climb” this landscape to the highest possible fitness—this is the most successful disease. But Vignuzzi suggests that a virus could be artificially altered to undergo exactly the wrong mutations, making it less fit and causing it to die off. Exactly how to do this remains a mystery, but it’s an exciting possibility.
Barbara Han, of the Cary Institute of Ecosystem Studies, presented her research on machine learning for forecasting zoonotic disease. Han takes a macro-ecological approach to disease, focusing on hosts. Factors like biodiversity and population density affect disease rates, so understanding zoonotic diseases means collecting a great deal of information about the animals that carry them. This information tends to be collected based on specific concerns about animal reservoirs; Han noted that since bats are a suspected reservoir for Ebola and other diseases, there’s been a massive surge in surveillance. It turns out, though, that they carry fewer zoonoses than we might expect. Right now, Han is studying bats to try to identify instances where viruses might spill back into bat reservoirs from human populations, making outbreaks harder to stop. She is also working with data about the health of rodent populations, with the hypothesis that lower biodiversity in a particular area will put humans at a higher risk for a spillover.
David O’Connor, from the University of Wisconsin-Madison, is looking at viruses that aren’t on the radar yet, though maybe they should be. O’Connor examines animal species to find traits that make spillover events likely. Specifically, he presented the theory that simian arteriviruses might be to blame for the mysterious simian hemorrhagic fever. There’s not enough information to know for sure without another outbreak, but O’Connor argues that there is enough information at our disposal to begin to make predictions “to the left of the surveillance curve,” and target surveillance at diseases that aren’t yet a top threat, but could emerge as one.
2 thoughts on “Predicting Emergence by Understanding the Past: Methods that Move Us towards Predictive Biology”