Novel coronavirus pneumonia, most of which affects human beings, is caused by human viruses, which are caused by viruses from other animal species, according to Spain. Therefore, early identification of high-risk viruses is helpful to prevent epidemics and epidemiological monitoring. We are talking about a huge challenge because there are about 1.6 million unknown viruses in wildlife around the world.
A new study just published in the Journal of Public Library of Science Biology in the United States shows that machine learning, that is, an artificial intelligence, can predict the probability of causing zoonosis according to the genome information of the virus. To develop relevant models, the researchers collected data sets of 861 viruses from 36 families. The machine learning model determines the probability of human infection based on the patterns observed in the viral genome. The researchers then used the best performing model to predict the pattern of potential zoonosis caused by other viruses sampled from various animals.
The researchers found that the viral genome may have generalizable features independent of the virus classification relationship, and may involve the pre adaptation process of virus infection in humans. In addition, they have developed a machine learning model for identifying potential zoonotic diseases using the virus genome.
It is reported that computer models are only a preliminary tool to identify zoonotic viruses with the potential to infect humans. They represent the first screening: confirmatory laboratory tests need to be carried out before judging the potential risk of viruses marked by these models. In addition, although these models predict whether the virus may infect humans, the ability to infect is only one of the components of the broader risk of zoonosis, which is also affected by factors such as the toxicity of the virus in humans, the ability of human transmission and the ecological conditions of human contact.
“Our results show that whether the virus has the potential to cause zoonosis can be inferred from its genome sequence. By highlighting the viruses most likely to cause zoonosis, genome-based classification can more effectively target further ecological and virological characteristics,” the researchers pointed out
Simon babayan, a researcher at the University of Glasgow, said: “These findings add a key part to the large amount of information we can extract from the virus gene sequence using artificial intelligence technology. The genome sequence is the first and usually the only information we have about the newly discovered virus.
The more information we can extract, the earlier we can determine the origin of the virus and the risk of zoonosis “As more viruses are ‘characterized’, our machine learning model will become more effective in identifying rare viruses. These viruses must be closely monitored and priority must be given to developing preventive vaccines against them.”