Instead of 10 years, it took 19 days.
That’s the kind of time saved by artificial intelligence software developed by researchers at the Vancouver Prostate Centre to find potential anti-viral drugs to treat people with COVID-19.
The AI program allowed lead investigator Dr. Artem Cherkasov and his team to sift through 40 billion molecular structures to find about 1,000 with the potential to target SARS-CoV-2, the virus that causes COVID-19.
They have identified seven possible candidates, and are now preparing to go to animal testing.
Dr. Cherkasov’s goal is to have an anti-viral compound ready sometime next year.
Advances in machine learning, he said, not only make finding new drugs hundreds of times faster, they also make it much cheaper and available to more people.
“Basically, we’re now looking to democratize certain components of drug discovery,” said Cherkasov, a senior scientist with the Vancouver Coastal Health Research Institute and professor in the Department of Urologic Sciences at the University of B.C.
“Any lab or centre, you name it, could have access to pretty massive drug discovery computing. Previously, you really had to have super-computing facilities to do this type of massive prediction.
“Now, any grad student from any place in the world can access pretty industrial-size computing power because of this.”
A paper on Cherkasov’s research is being published Tuesday by Chemical Science, one of the journals of the Royal Society of Chemistry.
The computer-aided drug discovery software is called Deep Docking, which learns as data is processed. Designed by Cherkasov, it builds on software he has been working with for more than 20 years.
In 2015, researchers Cherkasov and Paul Rennie designed a candidate drug to treat prostate cancer that Roche Pharmaceuticals paid $141.7 million in milestone payments through the regulatory process.
In that case, earlier iterations of software reviewed three million molecules to find a prostate drug prototype.
“Now we go through not three million but 40 billion molecules,” he said. “To make that leap, we used AI. AI allowed us to do things hundreds of times more efficiently.”
For a decade, Cherkasov had been working to find anti-virals for prostate cancer, but shifted to COVID-19 early in 2020 when it became clear that something dramatic was happening around the world with the newly discovered coronavirus.
Until now, most of the international attention has been on the spike protein of SARS-Cov-2, which allows the virus to infect cells.
Cherkasov and his team used Deep Docking to target something different — the virus’s main protease called Mpro, an enzyme important for replication.
He compared the reason for targeting the main protease as similar to the reason for the development of the “HIV cocktail” — a combination of three or more antivirals that helps to stop the human immunodeficiency virus from replicating and leading to AIDS.
“We’re hitting a different enzyme in the virus,” he said about SARS-CoV-2. “We’re hitting in a very different way. We need to hit the virus from multiple directions.”
What he is aiming for is to find the perfect chemical “key” to stop the COVID-19 virus.
“The lock is the enzyme,” he said. “We’re inserting a key inside the enzyme and basically breaking it. That’s the goal.”
Data collected by the team will be publicly accessible through Github and the Democratizing Drug Discovery with Deep Docking (D5) platform.