The jump of SARS-CoV-2 from animals to humans It caught us totally off guard. So much so that to this day we are not even clear about the animal that first infected a person. It appears that the virus arose in bats, but it is not known for sure what the kind of transition. And if it happened once, something similar may happen again in the future. For this reason, a team of scientists from the University of Glasgow has designed an algorithm of artificial intelligence, whose objective is precisely to try to predict these jumps.
Obviously, it cannot be known for sure. However, they hope that with him they can at least have some candidates to which to pay special attention, in order to prevent future catastrophes.
The procedure by which they have programmed and trained this artificial intelligence algorithm is described in the study just published in PLOS Biology. But what else does it count on?
The next virus that will jump from animals to humans
It is estimated that only a minority of 1.67 million species of animal viruses that exist is capable of infecting humans.
Diseases from pathogens that have passed from animals to humans are called zoonotic diseases.
Detecting them in time is very important, since the jump from animals to humans is a first encounter with the pathogen which can be very dangerous. After all, unless something known as cross immunity, which occurs when two viruses are very similar to each other, generally the fact that our immune system has never been exposed to them makes us very vulnerable. And this is how the epidemics or pandemics of the known as zoonotic diseases; among which, of course, is COVID-19.
There are many factors that influence whether this jump can be generated, but undoubtedly among them those related to the virus genome. And this is where these scientists thought artificial intelligence could be effective. You just had to teach the machine what genetic traits make a virus susceptible to passing from animals to humans. Thus, by introducing a new genetic sequence to it, you could determine if it is a virus that we should be concerned about.
Artificial intelligence to reduce search
To develop the models of machine learning on which this artificial intelligence is based, the study authors took a set of genetic data from 861 virus species, distributed in 36 families.
Artificial intelligence cannot determine the total risk, which also depends on factors such as virulence or the ability to be transmitted between humans
They then analyzed them, both based on their biological classification, as of your similar with viral species already known to infect humans. This allows the algorithm to determine the spots to detect when introducing a new genome.
When testing it the results were quite promising, although only what is known as candidate zoonoses were obtained. That is, the viruses that could perhaps jump from animals to humans were pointed out, but taking into account that it is not something that is going to happen for sure, since there are other factors involved, which could only be analyzed in the laboratory. Furthermore, artificial intelligence does not determine the risk total, which will also depend on the capacity of spread between humans, the virulence of the pathogen or even the ecological conditions at the time of exposure.
Ultimately, all of this would be a first step. Extensive and expensive laboratory tests are required to more reliably detect which viruses can jump from animals to humans. So narrowing down your search is always a good idea. And that’s what you get with this artificial intelligence. There is still a long way to go ahead of her, but the first step is especially important.