Newly developed technology has led to the discovery of more than a hundred thousand new types of viruses. This is what researchers are currently reporting in the specialist magazine “Nature”.
Tiny and widespread: viruses can be found everywhere, from glacial ice to the depths of the earth’s crust and even the atmosphere. Even we humans carry billions of viral species within us. Nevertheless, it is difficult to detect and study the viruses – a large part of the virus world remains hidden from science.
But now a team of researchers has discovered more than 160,000 previously unknown RNA viruses – more than ever before. They have artificial intelligence (AI) to thank for this.
New RNA viruses in particular have been difficult to detect so far. The reason: These viruses mutate quickly and have immense genetic diversity, which makes them difficult to classify. Traditional methods specifically searched for characteristic RNA building blocks, but this is very time-consuming and error-prone.
A team led by Xin Hou from the State Laboratory for Biocontrol in Shenzhen (China) developed an AI that made it possible to detect new types of viruses. The system, called Lucaprot, is based on a learning transformer model – similar to ChatGPT – that is specifically designed to detect viral signatures in RNA data.
For their study, Hou and his team trained Lucaprot with about 5,000 known RNA signatures and then had it analyze 51 terabytes of RNA data from environmental samples. These samples came from 1,612 locations worldwide and spanned 32 different habitats – from deep-sea sediments to Antarctic ice and hot springs.
The result was impressive: the AI revealed 161,979 new types of RNA viruses. “Finding so many new viruses in one fell swoop is mind-boggling,” said Edwards Holmes of the University of Sydney, one of the study’s lead authors. According to the study, this discovery expands the previously known virosphere (editor’s note: totality of all viruses existing on Earth) by one and a half times.
The newly identified viruses were therefore distributed across all ecosystems examined – the highest diversity was in wetlands, inland waters and wastewater. A particularly large number of new viruses have been found in Antarctic and marine sediments as well as in some inland waters.
But despite this large number of newly discovered viruses, researchers say we are only just scratching the surface: “There are still millions more viruses to discover,” emphasized Holmes. There are also large gaps in the knowledge about the evolution and ecology of these new types of viruses – for example which hosts they prefer to infect.
“The majority of RNA viruses known to date infect eukaryotes,” the scientists explained. In addition to humans, this also includes animals, plants and cell nucleus-bearing single-celled organisms. However, it is suspected that bacteria, for example, could also serve as hosts – this still needs to be researched.
“The next step is to train our AI to detect even more of this amazing diversity,” concluded Holmes, adding, “Who knows what surprises await us.”