Thanks to AI, we now have a shorter list of animals that may carry viruses threatening humans and therefore need further study.

About two-thirds of infectious disease agents come from animals. However, when a new virus emerges, we often don’t know exactly which animal it jumped from – just think of the current coronavirus that is on the move. It means that we are still too often behind the times. But now researchers in a new study We have shown that with the help of artificial intelligence we can better predict which viruses can infect humans, which animals they house and where they are likely to appear.

on the radar

Ideally, researchers would like to get viruses that have the potential (in the long term) to jump from animals to humans on the radar. But that’s easier said than done. “We often don’t know where a virus comes from,” researcher Daniel Becker told Scientias.nl. “This makes it difficult to control the spread of infection. At the same time, we need to know where to look for future zoonotic viruses. Because then we know which animals to monitor and where to take action.”

Thousands of mammal species

However, there are thousands of mammal species. And we simply have neither the time nor the resources to sample them all, piece by piece. “So we have to find a way to find out where the priority lies,” emphasizes Becker. “Model studies like ours are the first step in identifying which groups of animals are most likely to carry a particular dangerous virus. That way we can rigorously narrow down the list of animals harboring potentially risky viruses. We then also know better which specific viruses we need to characterize and when, where and how they can transfer to humans.”

Successor to SARS-COV-2

If we subsequently know better which animals are carriers of dangerous viruses and keep a close eye on them, we may also be better able to discover the successor to SARS-COV-2 in time. And that is very important if we want to prevent another pandemic. “We’ve known for some time that most of the new infectious diseases come from wildlife,” says Becker. “It is therefore likely that SARS-COV-2 will not be our last experience with a zoonosis. Whether ‘the successor’ will eventually trigger a pandemic on a similar scale to COVID-19 is harder to say for sure. But we can expect future spread of viruses to humans; especially with the increasing disturbance of the living environment and climate change.”

However, if you want to find these viruses, you have to start by profiling their hosts, such as their ecology and evolution. And with the help of artificial intelligence, researchers can then turn that data into concrete predictions about where to look for the next SARS-COV-2.

Betacoronaviruses

To gain more clarity on this, the research team used artificial intelligence to find out in the first quarter of 2020 which species of animals harbor so-called ‘beta coronaviruses’; a large group of SARS-like viruses, including the virus that caused the SARS outbreak between 2002 and 2004 and the virus that now causes COVID-19. A dangerous group of viruses. The researchers come to an interesting conclusion. Because the results show that there are more than 400 bat species worldwide that can be undetected hosts of beta-coronaviruses.

This photo shows a Rhinolophus rouxii; a mammal of the horseshoe nose family. According to the researchers, this bat is a likely and as yet undetected carrier of beta-coronaviruses. Image: Brock and Sherri Fenton

The new models confirmed some known virus reservoirs. For example, the results show that 21 species of horseshoe bats – a family of bats found in Europe, Africa, Asia, Australia and North America – carry beta-coronaviruses.

unnoticed

But the researchers also found some unexpected bat species. “Using the models, we found that certain bats – including Pugs and Chaerephon – most likely harbor undiscovered beta-coronaviruses,” says Becker. “We also identified geographic areas where the likely, but not yet sampled, beta-coronavirus hosts reside. For example, they occur not only in mainland Southeast Asia, but possibly also in the Malay archipelago, in sub-Saharan Africa and to a lesser extent in Europe and parts of the Americas.”

Shorter list

Thanks to AI, we now have a shorter list of animals that may carry viruses threatening humans and therefore need further study. “Now that we have identified these most likely hosts, the next step is to better monitor them and understand where and when beta-coronaviruses can jump,” said Becker. Scientists from around the world are now working together to test bats for coronaviruses based on the predictions from the current study. And that is an important step forward. “If we don’t have to spend as much money, resources and time looking for these viruses, we can put it into things that actually save lives,” said study researcher Colin Carlson. “We can invest in developing universal vaccines to tackle all those viruses at the same time. We can also prevent these viruses from spreading to people who live near bats.”

Artificial intelligence

It means that artificial intelligence can be an important weapon in the fight against the emergence of new dangerous viruses. “One of our main implications is that ‘trait-based models’—that is, models that use species characteristics to distinguish beta-coronavirus bats from other bat species—are much better results than other model types,” says Becker. “This may sound logical (the more we learn about species, the better we can distinguish which species might harbor viruses), but surprisingly this has not been demonstrated before. We now want to investigate further with which specific information we can improve our models. For example, we already use geographic data, the evolutionary history of bats and data on their body size. But the more we learn about bat immunology, genetics and physiology, the more reliable the results will become.”

Scientists can then try to use this information to prevent viruses from spreading to humans and nip another pandemic in the bud. “It’s important to note that beta-coronaviruses often don’t jump directly from bats to humans,” says Becker. “In both SARS-CoV and MERS-CoV this was done by means of an intermediate host; in SARS-CoV, for example, these were civet cats. When a virus jumps, it is often the result of environmental changes or human contact with wild animals. And by tackling those issues, we can also rigorously reduce the risk of infectious diseases.”