Deep Learning Tool May Accelerate COVID-19 Drug Discovery

A deep learning tool can offer more information about SARS-CoV-2 proteins to accelerate COVID-19 drug discovery, according to a study published in Chemical Science.

For more coronavirus updates, visit our resource page, updated twice daily by Xtelligent Healthcare Media.

Researchers from Michigan State University (MSU) Foundation repurposed deep learning models to focus on a specific SARS-CoV-2 protein called its main protease. The main protease is a cog in the virus’s protein machinery that’s critical to how the pathogen makes copies of itself. Drugs that disable the main protease could stop the virus from replicating.

The main protease is distinct from all known human proteases, which isn’t always the case. Drugs that attack the viral protease are therefore less likely to disrupt people’s natural biochemistry.

The SARS-CoV-2 main protease is also almost identical to that of the virus responsible for the 2003 SARS outbreak. This means that drug developers already have information about the main protease and chemical compounds called protease inhibitors that interfere with the protein’s function.

However, researchers still have gaps in understanding where these protease inhibitors latch onto the viral protein and how tightly. The MSU team used its deep learning models to predict these details for over 100 known protease inhibitors. The data enabled the team to rank those inhibitors and highlight the most promising ones, which can be very valuable information for labs and drug developers.

“In the early days of a drug discovery campaign, you might have 1,000 candidates,” said Guowei Wei, a professor in the Departments of Mathematics and Electrical and Computer Engineering.

The researchers noted that typically, all those candidates would move to preclinical tests in animals, and then the most promising ten or so can safely advance to clinical trials in humans. By focusing on drugs that are most attracted to the protease’s most vulnerable spots, drug developers can narrow down that list of 1,000 from the start, saving both time and money.

“This is a way to help drug developers prioritize. They don’t have to waste resources to check every single candidate,” Wei said.

However, the researchers noted that the models don’t replace the need for experimental validation, preclinical, or clinical trials. Drug developers still need to demonstrate that their products are safe before providing them for patients, which can take a number of years.

Researchers stated that because of this, antibody treatments will likely be the first therapies approved during the pandemic. But these antibodies target the virus’s spike protein rather than its main protease. The team expects that their deep learning models can help drug developers create COVID-19 therapies.

“If developers want to design a new set of drugs, we’ve shown basically what they need to do,” Wei said.

Before the pandemic, the MSU team was already developing machine learning models to help save drug developers time and money. The group trains these deep learning models datasets about proteins that drug developers want to target with therapeutics. The models can then make predictions about unknown quantities of interest to help guide drug design and testing.

“We knew this was going to be bad. China shut down an entire city with ten million people. We had a technique at hand, and we knew this was important,” said Wei.

This model adds to the MSU researchers’ work to apply advanced analytics to the current healthcare crisis. In August, the team developed a machine learning tool that analyzed mutations to the SARS-CoV-2 genome that have made the virus more infectious.

“Knowledge about the infectivity of SARS-CoV-2 is a vital factor for preventive measurements against COVID-19 and reopening the global economy,” Wei said at the time. “A crucial question is what are the ramifications of these mutations to COVID-19 transmission, diagnostics, prevention and treatment.”