Which viruses should we fear and fight tooth and nail to prevent another pandemic? Scientists often find it difficult to estimate, but artificial intelligence is now offering help.
Many emerging infectious diseases that affect people – such as COVID-19 – are so-called zoonoses. These are pathogens that originate in other animal species, but at some point jump to humans.
Difficult
Ideally, researchers would like to get viruses that have the potential (in time) to jump from animals to humans on the radar in time. But that’s easier said than done. It’s like looking for a needle in a haystack; of the millions of viruses harboring animals, only a few can infect humans as well. And when researchers discover a new virus among animals, it is also very difficult to quickly assess whether it has the potential to make the transition from animals to humans and therefore also to determine whether this virus has been studied further and closely in the field. must be kept in mind.
New research
But a new study, published in the magazine PLOS Biology could make the lives of researchers chasing zoonoses a little easier. In the study, scientists present a method based on the genome of a virus – often the only thing we know about newly discovered or poorly characterized viruses – to determine whether it is able to jump from animals to humans.
Artificial intelligence
The researchers make use of machine learning (a form of artificial intelligence). First, they compiled a database containing information about more than 800 types of viruses. They then built multiple artificially intelligent models that determined the likelihood of a virus infecting humans based on its evolutionary relationship to other viruses. Next, the researchers used the best-performing model to look for genetic aspects that viruses that the model believed could start infecting humans had in common. It shows that “a surprisingly large part of the zoonotic potential of viruses can be read from their genome sequence,” the researchers say. And then they developed an artificially intelligent model that can identify candidate zoonoses purely on the basis of the viral genome.
Promising
The promise of the model was shown when researchers ran it on a large number of viruses, including SARS-CoV-2. With no knowledge of previous SARS outbreaks, the model identified SARS-CoV-2 and closely related relatives as high-risk viruses. That is, the model estimates the chance that they can jump from animals to humans is quite high.
Cross
The model can therefore predict whether a virus is capable of infecting humans. But that’s all. For example, it cannot predict when the virus will spread, what symptoms people will develop when they catch the virus, and whether the virus will also be transmissible from person to person after it has spread to humans (and thus has the potential to cause a pandemic. does not offer us the successor to SARS-CoV-2 on a silver platter, but gives us a first step.By predicting which (newly discovered) viruses have the potential to cause misery among humans, the model can show researchers which viruses require urgent further investigation and close monitoring.
“Finding out what new viruses are capable of and how we should respond to them requires extensive characterization of the viruses, both in the lab and in their natural environment,” said researcher Nardus Mollentze. But because new viruses are being discovered at a rapid pace and characterizing them takes a lot of time, characterization is lagging behind considerably and there is a chance that we will only realize that they are dangerous long after viruses have been discovered. “When viruses are discovered, their genome sequence is often all we have, so developing an accurate machine learning method based on information trapped in that genome should allow us to better understand which animal viruses pose the greatest risk and should therefore be the first to be characterized and investigated.”
And hopefully this will also help us better prepare for future pandemics or possibly even prevent them. “Since most emerging human infectious diseases are caused by a small number of viruses that originated in other animal species, figuring out where to look for the next virus that could cause an epidemic remains a huge challenge,” said researcher Simon Babayan. “Now we have a fast, low-cost approach that allows us to specifically monitor and characterize those viruses that could potentially infect humans. And that, in turn, can help prepare for epidemics and pandemics in the future.”
Source material:
“Machine learning can predict which animal viruses risk infecting humans” – University of Glasgow
Image at the top of this article: PIRO4D (via Pixabay)