Colin Carlson, a biologist at Georgetown University, began to worry about measles.
The virus, discovered in 1930, spread among mice, killing them with relentless efficiency. But scientists have never considered it a potential threat to humans. Now Dr. Carlson, his colleagues and their computers are less secure.
Using a technique known as machine learning, researchers have spent the past few years programming computers to train for viruses that can infect human cells. Computers have combed through vast amounts of information about the biology and ecology of the animal hosts of these viruses, as well as the genomes and other characteristics of the viruses themselves. Over time, computers began to recognize certain factors that would predict whether a virus had the potential to spread to humans.
After computers proved their power on viruses that scientists had already studied extensively, Dr. Carlson and his colleagues placed them in the unknown, eventually creating a short list of animal viruses with the potential to cross the species barrier and cause outbreaks in humans. .
In recent series, algorithms have unexpectedly put the mouse measles virus at the forefront of high-risk pathogens.
“Every time we launch this model, it goes super high,” said Dr. Carlson.
Puzzled, Dr. Carlson and his colleagues took root in the scientific literature. They came across documentation of a long-forgotten epidemic in 1987 in rural China. The students contracted an infection that caused a sore throat and inflammation of their arms and legs.
Years later, a team of scientists conducted tests on throat swabs that were collected during the epidemic and stored. These samples, as the group reported in 2012, contained mouse measles DNA. But their study did not attract attention, and a decade later, measles is still not considered a threat to humans.
If the computer programmed by Dr. Carlson and his colleagues is correct, the virus deserves a new look.
“It’s just crazy that it’s lost in the huge pile of things that public health needs to sift through,” he said. “It actually changes the way we think about this virus.”
Scientists have identified about 250 human diseases that occur when an animal virus crosses the species barrier. HIV jumped from chimpanzees, for example, and the new coronavirus originated from bats.
Ideally, scientists would like to recognize the next virus before it infects humans. But there are too many animal viruses for virologists to study. Scientists have identified more than 1,000 viruses in mammals, but this is probably a small fraction of the true number. Some researchers suspect that mammals carry tens of thousands of viruses, while others estimate the number at hundreds of thousands.
To identify potential new transfusions, researchers such as Dr. Carlson use computers to find hidden patterns in scientific data. Machines can focus on viruses that may be particularly likely to cause human disease, for example, and can also predict which animals are most likely to contain dangerous viruses that we do not yet know about.
Barbara Hahn, a disease ecologist at the Cary Institute of Ecosystem Studies in Millbrook, New York, is collaborating with Dr. Carlson. Credit … Pamela Freeman / Carrie Institute for Ecosystem Research
“You seem to have a new set of eyes,” said Barbara Hahn, a disease ecologist at the Cary Institute of Ecosystem Studies in Millbrook, New York, who is working with Dr. Carlson. “You just can’t see in as many dimensions as the model can.”
Dr. Hahn first encountered machine learning in 2010. Computer scientists have been developing the technique for decades and are beginning to create powerful tools with it. Nowadays, machine learning allows computers to spot fraudulent credit charges and recognize people’s faces.
But few researchers have applied machine learning to disease. Dr. Hahn wondered if he could use it to answer open-ended questions, such as why less than 10 percent of rodent species contain pathogens known to infect humans.
It provides computer information on various rodent species from an online database – from their weaning age to their population density. The computer then looked for characteristics of rodents that are known to contain a large number of pathogens that skip species.
Once the computer has created a model, she tests it against another group of rodent species, seeing how well she can guess which ones are loaded with pathogens. In the end, the computer model reached an accuracy of 90 percent.
Dr. Hahn then turned to rodents, which have yet to be tested for pathogens, and compiled a list of high-priority species. Dr. Hahn and her colleagues predicted that species such as the mountain vole and the northern grasshopper mouse in western North America would be particularly likely to transmit alarming pathogens.
Of all the features that Dr. Hahn and her colleagues provided on their computer, the most important was the lifespan of rodents. Species that die young are carriers of more pathogens, perhaps because evolution invests more of its resources in reproduction than in building a strong immune system.
These results include years of diligent research in which Dr. Hahn and her colleagues browse environmental databases and research for useful data. More recently, researchers have accelerated this work by building databases specifically designed to train computers about viruses and their hosts.
The northern grasshopper mouse, one of the species that Dr. Hahn’s team predicts will carry an alarming pathogen. Credit … Rick and Nora Bowers / Alami
In March, for example, Dr. Carlson and colleagues unveiled an open access database called VIRION, which has collected half a million pieces of information on 9,521 viruses and their 3,692 animal hosts – and is still growing.
Databases like VIRION now allow more focused questions to be asked about new pandemics. When the Covid pandemic struck, it soon became clear that it was caused by a new virus called SARS-CoV-2. Dr. Carlson, Dr. Hahn, and their colleagues have developed programs to identify animals that are most likely to host relatives of the new coronavirus.
SARS-CoV-2 belongs to a group of species called beta-coronaviruses, which includes the viruses that caused the SARS and MERS epidemics in humans. Betacoronaviruses infect bats for the most part. When SARS-CoV-2 was discovered in January 2020, 79 species of bats were known to carry them.
But scientists have not systematically searched for all 1,447 species of beta-coronavirus bats, and it will take many years to complete such a project.
By submitting biological data on different species of bats – their diet, wing length, etc. – In their computer, Dr. Carlson, Dr. Hahn and their colleagues have created a model that can predict bats most likely to contain beta-coronaviruses. They found more than 300 species that match the bill.
Following this prediction, in 2020, researchers did discover beta-coronaviruses in 47 species of bats – all of which were on the lists of predictions created by some of the computer models they created for their study.
Daniel Becker, a disease ecologist at the University of Oklahoma who also worked on beta-coronavirus research, said it was amazing how simple characteristics such as body size could lead to powerful predictions about viruses. “Much of this is the low-hanging fruit of comparative biology,” he said.
Now Dr. Becker continues from his own backyard to the list of potential hosts for beta-coronavirus. It turns out that some bats in Oklahoma are expected to be sheltered.
If Dr. Becker finds beta-coronavirus in the backyard, he will not be able to say right away that it is an immediate threat to humans. Scientists must first conduct diligent experiments to assess the risk.
Dr. Pranaw Pandit, an epidemiologist at the University of California, Davis, warns that many of these models are in the works. When tested on well-studied viruses, they do much better than accident, but they could do better.
“It’s not at a stage where we can just take these results and create a signal to start telling the world, ‘This is a zoonotic virus,'” he said.
Nardus Molenze, a computer virologist at the University of Glasgow, and his colleagues are pioneers in a method that can significantly increase the accuracy of models. Instead of looking at the virus’s hosts, their models look at its genes. The computer can be taught to recognize subtle features in the genes of viruses that can infect humans.
In their first report on this technique, Dr. Molenze and his colleagues developed a model that can correctly detect viruses that infect humans in more than 70 percent of cases. Dr. Molenze still can’t say why his gene-based model works, but he has some ideas. Our cells can recognize foreign genes and send an alarm to the immune system. Viruses that can infect our cells may have the ability to mimic our own DNA as a form of viral camouflage.
When they applied the model to animal viruses, they came up with a list of 272 species at high risk of spread. This is too much for virologists to study in depth.
“You can only work on so many viruses,” said Amy de Witt, a virologist at Rocky Mountain Laboratories in Hamilton, Montana, who is monitoring research on the new coronavirus, flu and other viruses. “For our part, we will really have to limit it.”
Dr. Molenze acknowledged that he and his colleagues need to find a way to determine the worst of the worst among animal viruses. “This is just the beginning,” he said.
To continue his initial research, Dr. Molenze is working with Dr. Carlson and colleagues to combine data on virus genes with data related to the biology and ecology of their hosts. Researchers are getting some promising results from this …
Add Comment