BEIJING — This year’s Nobel Prize results mean that artificial intelligence (AI) technology is not only a leading trend in computer science, but is also having a growing impact on disciplines such as biology and chemistry. It offers scientists a new research approach: using AI to unlock the secrets of nature.

One of the latest examples is related to virology. An international research team has used artificial intelligence technology to detect hundreds of thousands of RNA viruses in global ecosystems, demonstrating the enormous potential of artificial intelligence algorithms in virus detection and blazing new paths for virology.

A team of researchers from Sun Yat-sen University School of Medicine, as well as Zhejiang University, Guangzhou University, the University of Sydney and other institutions conducted the study, reporting the discovery of 180 supergroups of RNA viruses and more than 160,000 global RNA virus species. .

The study, which was recently published in the journal Cell, is the largest study of RNA viruses to date, significantly expanding knowledge of global RNA viruses.

NEW AI ALGORITHM

Viruses are an important component of Earth’s ecosystems and are closely linked to human health. However, the number of known virus species is still very limited. Scientists can use gene sequencing technology to compare the similarity of unknown viruses to known viral nucleic acid sequences, thereby identifying new viruses.

However, this method is based on existing knowledge about viruses. When studying RNA viruses that are highly divergent, numerous, and prone to mutation, sequence homology comparison cannot work effectively.

Researchers have proposed a new solution using artificial intelligence technology. According to Shi Mang of Sun Yat-sen University School of Medicine, who is also one of the authors of the research paper, AI algorithm models can detect viruses that were previously overlooked or not even known.

“During epidemics, the speed and accuracy of artificial intelligence technologies can help scientists quickly identify potential pathogens,” Shi said.

He led the team that used the core algorithm for the research, called LucaProt, a deep learning Transformer model. After extensive exploration of viral and non-viral genomic sequences, it can autonomously generate a set of virus identification criteria to find viral sequences from large RNA sequencing data sets.

NEW TYPES OF RNA VIRUS

According to the study, LucaProt demonstrated high accuracy and specificity, with a false-positive rate of 0.014 percent and a false-negative rate of 1.72 percent.

The team conducted viral searches of 10,487 RNA sequencing data from global environmental biological samples and discovered more than 510,000 viral genomes, representing more than 160,000 potential viral species and 180 RNA virus supergroups.

Among them, 23 supergroups could not be identified by traditional sequence homology methods. They can be called the “dark matter” of the viral community.

The study found that these viruses are common in various environmental environments on Earth. The greatest diversity of viruses is found in leaf litter, wetlands, fresh water and waste water. Significant diversity and abundance of viruses are also found in extreme environments such as Antarctic sediments, deep-sea hydrothermal vents, activated sludge, and saline-alkali barrens.

According to Hou Xin, the first author of the paper, these viruses include not only pathogens that infect humans, but also those that exist in the environment and infect various organisms. They can infect various animals, plants, single-celled protozoa, fungi and bacteria.

“A deeper understanding of viruses in the environment can help us better understand how the entire ecosystem works. Moreover, we can use this method to detect viruses closely related to human diseases for surveillance and early warning of emerging diseases,” Hou said.

“The traditional classification system has become inadequate for new viruses, the diversity of which far exceeds human imagination. What we are seeing now is just the tip of the iceberg,” Shi said.

NEW TOOL FOR MORE RESEARCH

It is a model specifically designed for detecting RNA viruses, but it also includes the ability to recognize protein sequences and implicit structural information and can be used to determine protein functions.

According to the study, the LucaProt model helped researchers identify genomic structures beyond previous knowledge of viruses, revealing the flexibility of RNA virus genomic evolution.

The study also identified many viral functional proteins, especially those associated with bacteria, indicating that there are more types of RNA bacteriophages, viruses that infect bacteria, to be studied.

The research team open-sourced the model and shared it with scientists around the world online.

Li Zhaorong of Alibaba Cloud Intelligence’s Apsara Lab, another corresponding author, believes that AI is gradually changing the way scientists approach various scientific problems.

“This model is becoming a cutting-edge tool in identifying viruses and is also being applied to other types of protein identification and discovery of their functions,” Li said.

Xu Jianguo, an academician at the Chinese Academy of Engineering, said LucaProt’s success marks a breakthrough for artificial intelligence algorithms in virus detection. In the future, AI is expected to become a major tool in microbiology and can be used to predict the pathogenicity of viruses to humans.