COVID-19: Could AI and supercomputers unlock the pathogen puzzle?

In partnership with The European Commission
COVID-19: Could AI and supercomputers unlock the pathogen puzzle?
Copyright euronews
Copyright euronews
By Julian GOMEZ
Share this article
Share this articleClose Button
Copy/paste the article video embed link below:Copy to clipboardCopied

Futuris looks at how Artificial Intelligence and High Performance Computing are helping in the fight against COVID-19.

Futuris looks at how Artificial Intelligence and High Performance Computing are helping in the fight against COVID-19.

"How and to what extent can Artificial Intelligence and High Performance Computing help in the fight against Covid-19? In the hunt for answers, European scientists are combining algorithms, biochemistry and molecular screenings as part of a research project that’s seeking to identify potentially new molecules that could be used for drugs against the coronavirus.”
Julián López Gómez

"coronavirus has a far greater complexity"

Computer scientists at this laboratory are using massive computational power against the pandemic.They say the enemy they are fighting is unlike any other.

"Compared to other viruses that we have worked on, for instance the Zika virus, the coronavirus has a far greater complexity. It is composed of many more proteins, that have very different biological roles,”says Andrea Beccari, a computer scientist for bio-pharmaceutical company Dompé and Project Coordinator for the E4C Project.

The team taking part in this European research project are trying to identify molecules that can block the progression of the virus inside the human body. The problem is researchers are having to explore a huge library of around 500 billion molecules.

To identify the most promising candidates in the shortest possible time, the scientists have turned to supercomputers. These can process up to 3 million molecules per second.

"Computers are essential, as we’re able to work at the same time with all 25 proteins of the virus which are involved in the various mechanisms related to infection, replication and block the human immune system - everything in a simultaneous way," explains Beccari.

3D models of pathogen proteins

The search for new molecules against COVID-19 also requires super-computers to generate and analyse 3D models of the proteins of pandemic pathogens.

“We are able to identify molecules that could be useful to counteract the coronavirus, both amongst molecules that already exist - through the so-called repurposing process - and by using new libraries of molecules, applying models that we have already optimised,” says Carmine Talrico, a computational chemist at Dompé.

Hunting for molecules that block COVID-19

After identification, promising molecules are shipped to Leuven in Belgium. Researchers there first infect cells with the SARS-CoV-2 virus. They then add the candidate molecules and study the reaction of the infected cells.

This complex research is partially done in a unique automated facility that works under low pressure, so biological material can not escape. The potential of any given molecule can be assessed in 48 hours.

"These robotic arms are taking the plates in which we have got the cells and the virus. These arms put these plates in the pipeting machines. After some time they put, for instance, the virus on top of the cells, and then they put the mix in the incubator at 37°C where they will stay for a couple of days. And they, after a couple of days, the robotic arms are taking the plates out of the incubator, and they put them in the automated microscope that is reading and checking if some of the compounds are blocking the virus," says Johan Neyts, a professor of virology at KU Leuven.

Scientists say they hope their research will eventually contribute to the development of efficient new drugs to be able tackle COVID-19 in the medium-term.

Share this article

You might also like