Mutations under scrutiny: How realistic is an early warning system for virus variants?

What we wish for is one thing, namely the end of the pandemic.

Something completely different is what the pandemic allows.

Seen from the perspective of the virus: The machine is humming, and that almost never before.

Around three million infections are detected worldwide every day, most of them without symptoms, and more and more infections of immunocompromised people.

In the journal "Cell Host & Microbe", a research team led by the South African infectiologist Alex Sigal from the Max Planck Institute for Infection Biology in Berlin spent six months tracing the genetic evolution of the Sars-CoV-2 pathogen in an HIV patient.

Their conclusion: D190, as the multiply mutated virus was called, found an ideal incubator in the immunocompromised AIDS patient.

Joachim Müller-Jung

Editor in the feuilleton, responsible for the "Nature and Science" department.

• Follow I follow

But nobody got infected with D190, at least it didn’t trigger a wave.

Delta, on the other hand, is massive, and Omicron is even more massive.

Before these two virus variants began their worldwide triumph in September 2020, almost 300 new Sars-CoV-2 variants were registered in the GISAID genome database every week, a year later there were already 7,000, and by the end of December 2021 there were an estimated 12,000 weekly.

The incubator of mankind supplies supplies faster and faster.

At least a quarter of a million mutations are known today, but of course not all of them are dangerous. On the contrary, most mutations weaken the virus. Just: what are they? And above all: Can we predict which mutations in which combinations have the potential to be more infectious and even more insidious when it comes to outwitting our immune system?

In fact, many researchers around the world are now working on a kind of early warning system for virus variants.

Admittedly, none of them are fully developed, and some, like the computer model presented by Swiss biologists in the “Proceedings” of the American National Academy of Sciences, which “anticipates” new mutations on the basis of statistical analyses, seem all too theoretical.

Nevertheless, there is progress.

In "Science Translational Medicine", an American group determined the different binding properties of various genetic changes and calculated the "virus fitness" from them.

On average, it was possible to see four months in advance which mutations that were already known would spread.

Also, what mutations protect the virus from being tailored,

Despite enormously growing amounts of data, evolution cannot be predicted, and even less how combinations of many mutations – in the case of Omicron more than fifty – will affect it. Important here are mutations in the crucial binding molecule of the virus, the spike protein, of which more than 13,400 are known alone. Spike is the point of attack of the most effective antibodies and is therefore particularly sensitive. In a preprint on "bioRxiv", the Mainz-based immunologist and Biontech founder Ugur Sahin and his group with AI experts from the London company InstaDeep presented a mechanical early warning system for "high-risk variants" that focuses on spike mutations and their potential for immune escape. Laboratory experiments supplemented the AI ​​analyses. Conclusion:Out of 13 variants identified by the World Health Organization (WHO), twelve could be identified as potentially dangerous on the basis of genome data alone - mathematically two months before the WHO threshold value for variant spread of 1500 infections was exceeded. A few weeks' advance warning would be a win. However, such a real-time early warning system only works reliably if the virus genome data flows as seamlessly as possible. However, this is still lacking almost everywhere.However, such a real-time early warning system only works reliably if the virus genome data flows as seamlessly as possible. However, this is still lacking almost everywhere.However, such a real-time early warning system only works reliably if the virus genome data flows as seamlessly as possible. However, this is still lacking almost everywhere.