Happy Nice Mall

Colin Carlson, a biologist at Georgetown University, began to worry about mousepox.

Discovered in 1930, the virus spreads among mice, killing them. brutal efficiency. But scientists have never considered it a potential threat to humans. Now Dr. Carlson, his colleagues and their computers aren’t so sure.

Using a technique known as machine learning, researchers have spent the last few years programming computers to teach them about viruses that can infect human cells. Computers scanned vast amounts of information about the biology and ecology of these viruses’ animal hosts, as well as their genomes and other properties. Over time, computers began to recognize certain factors that could predict whether a virus had the potential to spread to humans.

After computers proved their mettle on viruses that scientists were already working hard on, Dr. Carlson and his colleagues placed them in the unknown and eventually compiled a short list of animal viruses that have the potential to bypass the species barrier and cause human epidemics.

In recent studies, algorithms unexpectedly placed the mousepox virus in the top ranks of risk pathogens.

Dr. “Every time we run this model, it goes super high,” Carlson said.

Confused, Dr. Carlson and his colleagues have taken root in the scientific literature. They stumbled upon a long-forgotten document epidemic In rural China in 1987. Schoolchildren came in with an infection that caused a sore throat and inflammation of their hands and feet.

Years later, a team of scientists ran tests on throat swabs that were collected and stored during the epidemic. These samples contained mousepox DNA, as the group reported in 2012. But their work has received little attention, and rat pox is still not considered a threat to humans, ten years later.

Dr. If the computer programmed by Carlson and his colleagues is correct, the virus deserves a new look.

“It’s crazy that this is getting lost among the sheer number of things public health needs to review,” he said. “This is actually changing the way we think about this virus.”

Scientists have identified about 250 human diseases that occur when an animal virus bypasses the species barrier. HIV jumped for example from chimpanzees and new coronavirus origin bats.

Ideally, scientists would like to recognize the next spreading virus before it starts infecting humans. But there are too many animal viruses that virologists can’t study. Scientists have detected more than 1,000 viruses in mammals, but this is likely a very small fraction of the real number. Some researchers suspect that mammals carry it. tens of thousands the number of viruses, others the number hundreds of thousands.

To identify potential new spreads, Dr. Researchers like Carlson use computers to detect hidden patterns in scientific data. The machines could, for example, focus on viruses that are particularly likely to cause a human disease, and also predict which animals are most likely to harbor dangerous viruses that we do not yet know.

D., a disease ecologist at the Cary Institute for Ecosystem Research in Millbrook, NY, and Dr. “It feels like you have a new pair of eyes,” said Barbara Han, who collaborated with Carlson. “You can’t see in as many dimensions as the model can see.”

Dr. Han first met with machine learning in 2010. Computer scientists had been developing this technique for decades and were starting to build powerful tools with it. nowadays, machine learning It allows computers to detect fake loan fees and recognize people’s faces.

But few researchers have applied machine learning to diseases. Dr. Han wondered if he could use it to answer open questions, such as why less than 10 percent of rodent species harbor pathogens known to infect humans.

It provided a computer with information about various rodent species from an online database – everything from weaning ages to population densities. The computer then looked for characteristics of rodents known to harbor pathogens that spread to numerous species.

Once the computer built a model, it tested it against another group of rodent species to see how well it could predict which ones were filled with disease-causing substances. Eventually, the model of the computer reached a value of accuracy. 90 percent.

Later Dr. Han turned to rodents that had not yet been studied for spreading pathogens and put together a list of high-priority species. Dr. Han and colleagues estimated that western North American species such as the mountain vole and northern locust vole were most likely to carry pathogens of particular concern.

Dr. Of all the features Han and colleagues provided to their computers, the most important was the rodent’s lifespan. Species that die young turn out to carry more pathogens, perhaps because evolution has devoted their resources to reproduction rather than building a strong immune system.

These results, Dr. It involved years of painstaking research, in which Han and colleagues combed through ecological databases and scientific studies looking for useful data. More recently, researchers have stepped up this work by creating databases explicitly designed to give computers information about viruses and their hosts.

For example, in March, Dr. Carlson and colleagues open An open-access database called VIRION, which collects half a million pieces of information on 9,521 viruses and their 3,692 animal hosts and continues to grow.

Databases like VIRION now make it possible to ask more focused questions about new pandemics. When the Covid pandemic hit, it was soon discovered to be caused by a new virus called SARS-CoV-2. Dr. Carlson, Dr. Han and colleagues created programs to identify the animals most likely to harbor relatives of the novel coronavirus.

SARS-CoV-2 belongs to a group of species called betacoronaviruses, which also includes viruses that cause epidemics of SARS and MERS among humans. Mostly, betacoronaviruses infect bats. When SARS-CoV-2 was discovered in January 2020, 79 species of bats were known to carry them.

But scientists have not systematically searched all 1,447 bat species for betacoronaviruses, and such a project would take many years to complete.

Dr. Carlson, Dr. By feeding biological data about various bat species (diet, length of wings, etc.) into their computers, Han and colleagues created a model that could offer predictions about bats that are likely to be found. to host betacoronaviruses. They found more than 300 species that fit the bill.

Since that prediction in 2020, researchers have indeed found betacoronaviruses in 47 bat species – all of which are on prediction lists generated by some computer models they’ve created for their study.

Daniel Becker, a disease ecologist at the University of Oklahoma, betacoronavirus study, said that it is striking that simple features such as body size can lead to strong predictions about viruses. “Most of them are the low-hanging fruit of comparative biology,” he said.

Dr. Becker is now tracking the list of potential betacoronavirus hosts from his own backyard. It is speculated that some bats in Oklahoma harbor them.

Dr. If Becker finds a betacoronavirus in the backyard, he won’t be able to immediately say it’s an imminent threat to humans. Scientists would first have to perform painstaking experiments to assess the risk.

Pranav Pandit, an epidemiologist at the University of California at Davis, warns that these models are a work in progress. When tested on well-studied viruses, they do significantly better than random chance, but they can do better.

“It’s not at a stage where we can take these results and create a warning to start saying ‘This is a zoonotic virus’ to the world,” he said.

Nardus Mollentze, a computational virologist at the University of Glasgow, and colleagues have pioneered a method that could significantly improve the accuracy of the models. Instead of looking at a virus’s hosts, their model looks at its genes. A computer can be taught to recognize subtle features in the genes of viruses that can infect humans.

in them first report On this technique, Dr. Mollentze and colleagues have developed a model that can accurately recognize viruses that infect humans more than 70 percent of the time. Dr. Mollentze can’t yet explain why his gene-based model works, but he has some ideas. Our cells can recognize foreign genes and send alarms to the immune system. Viruses that can infect our cells may have the ability to mimic our own DNA as a kind of viral camouflage.

Applying the model to animal viruses, they came up with a list of 272 species at high risk of spreading. This is too much for virologists to delve into.

“You can only study so many viruses,” said Emmie de Wit, a virologist at Rocky Mountain Laboratories in Hamilton, Mont., who oversees research on the novel coronavirus, flu, and other viruses. “On our part, we’re going to have to really narrow it down.”

Dr. Mollentze acknowledged that he and his colleagues needed to find a way to detect the worst of the worst among animal viruses. “This is just the beginning,” he said.

To follow up on his initial work, Dr. To combine data on the genes of viruses with data on the biology and ecology of their hosts, Dr. He works with Carlson and his colleagues. Researchers are seeing some promising results from this approach, including the exciting mousepox lead.

Other data types can make predictions even better. For example, one of the most important features of a virus is the coating of sugar molecules on its surface. Different viruses result in different sugar molecule patterns, and this arrangement can have a huge impact on their success. Some viruses can use this molecular freeze to hide from their host’s immune system. In other cases, the virus may use the sugar molecules to attach itself to new cells and trigger a new infection.

This month, Dr. Carlson and colleagues have published an online commentary claiming that machine learning can gain a lot of insights from the sugar coating of viruses and their hosts. Scientists have already gathered much of this information, but it has not yet been put into a form that computers can learn.

Dr. “According to our instincts, we know a lot more than we think,” Carlson said.

Dr. de Wit said machine learning models could one day guide virologists like him in studying specific animal viruses. “There will definitely be a huge benefit from this,” he said.

But he noted that so far the models have focused mainly on the potential of a pathogen to infect human cells. Before it can cause a new human disease, a virus must spread from one person to another and cause severe symptoms along the way. It is waiting for next-generation machine learning models that can also make these predictions.

“What we really want to know is not which viruses can infect people, but which viruses can cause an epidemic,” he said. “So this is the next step that we really need to understand.”