From a quantitative point of view, viruses are the true masters of the Earth : it is estimated that there may be up to ten quintillion – that is, 10 31 – of these very special microorganisms on our planet. Although some scientists deny them the category of living beings, since they are unable to survive outside their hosts, they have a terrifying capacity for expansion, change and destruction.
The problem with studying and census these elusive creatures is that they can hardly ever be grown in the laboratory. Furthermore, their small genomes evolve very fast. The only way to identify them is to extract samples from different environments and try to sequence their DNA by comparing it with the genetic material of other viruses or bacteria .
And here comes artificial intelligence: a study presented at a meeting of the US Department of Energy (DOE) announces that this emerging technological tool has managed to uncover 6,000 previously unknown virus species. Specifically, the researchers used the so-called machine learning or automatic learning, a digital system that manages to detect patterns in huge amounts of data autonomously after being properly trained.
What Simon Roux, a biologist at the Joint Genome Institute in Walnut Creek, California, and his colleagues did was train an artificial intelligence algorithm with two data packages : one contained 305 sequences from the genome of Inoviridae – a well-known family of viruses that infect bacteria – and the other, about 2,000 sequences of other microorganisms. Then they put the system to work with a barrage of data: 10,000 genomes of Inoviridae viruses, which it had to group into different species. Variation in DNA between some varieties suggests that perhaps what experts consider a single family – Inoviridae – are actually multiple.
Rummaging through the zoo compost
At the same meeting, another study was presented that applies the potential of machine learning to virology. Deyvid Amgarten applied his algorithm to find new species in compost from the Sao Paulo zoo in Brazil, analyzing the density of genes in DNA strands of a certain length. The ultimate goal of this research is to discover the proteins that viruses produce and to improve the production of organic compost or, in other words, of matter decomposed by the action of microbes.
Amgarten took advantage of the VirFinder tool, developed in 2017 by the team of Jie Ren, a computational biologist at the University of Southern California. VirFinder was able to find out, for example, which viruses were more abundant in the feces of cirrhosis patients compared to healthy people, to establish a link between the microorganisms and severe liver disease.
It is difficult to overstate the importance of the findings of VirFinder and other artificial intelligence systems applied to the study of viruses: one of the great challenges of medicine is to find out the role that these microorganisms play in the appearance of diseases as enigmatic as the syndrome of chronic fatigue and disorders of the immune system, as highlighted in the information in the journal Nature that accounts for the discoveries.
