The computer matches the virus with carriers

Through what kind of mammal could a new virus infect humans? Researchers at the University of Liverpool have developed a new machine learning system to predict this. The system discovered more than 20,000 new links between known viruses and potentially vectors, mammals.

According to recent estimates, mammals transmit thousands of types of viruses to each other and only 1 percent of the viral diversity in those animals is known. Depending on the type of virus, infectious diseases have a limited number or a large number of potential carriers. Immunodeficiency viruses such as HIV and human immunodeficiency virus (FIV), the viruses that cause AIDS and feline AIDS respectively, can infect a few species. The rabies virus that causes rabies and the SARS-CoV-2 coronavirus are different. It can be spread in many different animal species. They are also zoonotic and can therefore jump from animal to human.

The study showed that the number of links between known viruses transmitted from animals to humans and land mammals is five times higher than expected. Bats and rodents in particular are showing more connections to viruses that can be transmitted to humans. In addition, the system predicts a fivefold increase in the number of links between wild mammals and viruses in livestock and domestic animals. Lack of knowledge about certain mammalian species increases the risk that they will not be detected as potential vectors of viruses.

Image: Observed and predicted networks of interactions between wild and semi-domestic mammal hosts and known viruses. Credit: Maya Warda

The researchers are currently studying whether they can extend the computer model to predict transmission of the virus in ticks and insects. Scientists can then intervene more specifically in the event of a virus outbreak through contact between invertebrates, birds and mammals.

Source: University of Liverpool, UK