The virus cannot be confused with a single puncture. When it makes copies of itself, bugs – mutations – often creep in and sometimes work positively. As for the virus then. It becomes more infectious or escapes the immune system. A vaccine can play tricks on this. The virus that has changed shape is no longer recognized by the immune system. Or the antibodies no longer stop it.
So vaccine developers are looking for parts of the virus that can’t change so quickly. But how do you find it? You can test all known and possible mutants in the laboratory by releasing antibodies to them, but this is a huge task.
American scientists believe that a computer can do this faster. And definitely a self-learning computer system. But the biggest leap in thinking they made was that they didn’t have to build such a system. They already had a decent thing on the shelf: a language computer.
That was – and still is – challenging. Computers can generate sentences, but it is difficult to get a computer to say something meaningful. Anyone can express their thought in hundreds of ways, but you just have to change the sentence and it doesn’t make sense anymore. Computer systems have mastered this skill somewhat when presented with numerous texts and had to figure out for themselves when the sentence was grammatically correct and meaningful.
A virus is more than just a genetic code that describes how to make a copy of this virus. This symbol is bound by biological rules. You cannot make changes to it with impunity. This usually does not result in viable viruses. You can say: These biological bases are the rules of virology.
The immune system reads the virus
But what is the parallel with the meaning of the text? Immune system response. After all, the immune system reads the virus and then tries to fight it.
Researchers at the American Massachusetts Institute of Technology used language computer architecture to understand virus mutations. They have had their system study thousands of variants of the influenza virus, AIDS virus and the new coronavirus so it can teach itself which mutations are viable. What were the mutations that gave the virus different properties? For example, what makes a virus no longer contagious to a pig, but to humans? And on top of this: How can viruses evade an immune response? This system could accurately indicate the escape routes a virus has, as they write in Science.
It’s very smart that they were able to do this, says Berend Snel, professor of bioinformatics at Utrecht University. “Through so-called deep learning, you get much more information from the genetic code than the traditional method. It’s an important step to speed up the investigation. At the same time, vaccine developers will not blindly trust these predictions but want to confirm them in the laboratory.”
For most viruses, it is clear which parts are mutable and which are not. Snell: “Some viruses mutate so much that the genetic code does not appear to have a stable region. Using this approach, you can predict which mutations will allow escape from the immune system without affecting the virus’s viability. Which mutations are not important.
Fear of 2020: An Overview of the Virus
Coronavirus has now spread throughout the world, but has not yet revealed its identity.