Can a genetic model predict next year’s flu strain?
The seasonal flu has met its enemy, and it’s calculus.
A theoretical physicist and computational biologist analyzed the genetic code of thousands of strains of Influenza A that occurred over a 44-year period to create a model that accurately predicts which strain will prevail in the pitched evolutionary battle between human antibodies and the rapidly mutating virus.
Their method proved more accurate for selecting an appropriate vaccine than the current method used by public health officials, according to a report published online Wednesday in the journal Nature.
The researchers, from the University of Cologne in Germany and Columbia University, examined Influenza A/H3N2, a seasonal strain that causes thousands of deaths annually. They focused on the H part – short for hemagglutinin, a spike-shaped protein that seeks out sugars in human cells and binds to them, allowing the virus to inject its deadly code.
Human antibodies – naturally launched or gigged into action by vaccines - engage in a constant arms race with this wildly mutating protein, making flu vaccination season something of an educated guessing game. World health officials have been reasonably accurate in identifying which resistant strains are emerging as a new threat. But their method has a lot of uncertainty. A study in 2010, in fact, called several popular diagnostic methods “questionable.”
But inaccuracy also is at the heart of the survival for a virus, which shuffles its code enough to create variations that don’t show up on the radar of the human immune system.
“This is a really fast-evolving virus,” said study coauthor Marta Luksza, a computational biologist at Columbia University. “Individual strains are very short-lived. Very, very rarely are there two identical strains. They mutate, they infect another individual. But they do share common mutations still.”
Luksza and coauthor Michael Laessig estimated the frequency of these mutations in the viral RNA and factored in such variables as infection rates to come up with a model predicting a strain’s evolutionary fitness.
Since the researchers already knew the outcome of the evolutionary arms race from the record of flu strains from 1968 to 2012, they tested whether their model would have identified the fittest strains. It predicted the rise of the correct lineage in 93% of cases.
Perhaps more importantly, the model chose a more genetically matched strain than the one deployed as a vaccine.
Luksza cautioned that the study focused on one type of influenza. When they tested the model against an extinct H1N1 strain, results were less precise.
“In principle, there’s no reason why it shouldn’t be applicable to other strains,” Luksza said. But there may be complicating factors involved in the evolution of other strains that would have to be incorporated into the model, she added.
“I think our model could be used as guidance for the existing way of choosing vaccines,” Luksza said.
