It is estimated that there are around 1.7 million types of viruses that attack different types of animals. Some of them have or can develop the ability to attack humans. This applies to a few of them, but how do you know in advance which one?
This question was further investigated by British researchers at the University of Glasgow. They hope that in the future, artificial intelligence and machine learning can be used to defeat viruses.
What the researchers found is that, to a surprising degree, it appears that the genetics of a virus determine whether or not it can become zoonotic, that is, pass from animals to humans.
This is good news, as genetic sequencing is often the first and only source of information when newly discovered viruses emerge. Thus, we have a better chance than before to quickly determine the origin of the virus and to assess the animal risks it may present.
The more viruses that are tested and characterized in this way, the better the machine learning model will be.
“This could be of great help in identifying rare viruses that should be closely monitored and prioritized for the development of a preventive vaccine,” said Simon Papian, one of the researchers behind the project. phys.org.
not on the way
The research team began by collecting a dataset of 861 known viruses from a total of 36 different virus families. They then built various machine learning models that determined the likelihood of infecting humans, based on the classification and relationship of known viruses that can infect humans.
Then they used the model that gave the best result to analyze the probability that several viruses in animals could infect humans.
This provides a good basis for future laboratory research, but it is still only a step in the right direction. The method the research team is developing, for example, says nothing about how easily the virus is transmitted between humans, or how viruses are able to cause disease.