Data Scientists Use Machine Learning to Discover COVID-19 Treatments
March 25, 2020 – Using machine learning algorithms, two data scientists are working to quickly discover effective treatments for COVID-19.
Andrew Satz and Brett Averso, graduates of the Data Science Institute at Columbia University, have launched a startup called EVQLV that creates algorithms capable of computationally generating, screening, and optimizing hundreds of millions of therapeutic antibodies.
Using this technology, the pair is working to discover treatments that will most likely help individuals infected by the virus that causes COVID-19. The machine learning algorithms are able to rapidly screen for therapeutic antibodies with a high probability of success.
While conducting antibody discovery in a lab can typically take years, these algorithms can identify antibodies that can fight against the virus in just a week.
“We are reducing the time it takes to identify promising antibody candidates,” said Satz. “Studies show it takes an average of five years and a half billion dollars to discover and optimize antibodies in a lab. Our algorithms can significantly reduce that time and cost.”
After performing computational antibody discovery and optimization, EVQLV sends the promising antibody gene sequences to its laboratory partners. Laboratory technicians then engineer and test the antibodies, a process that takes a few months as opposed to several years. Antibodies found to be successful will move on to animal studies and finally human studies.
With the urgency surrounding coronavirus and the worldwide effort to fight it, the pair believes it may be possible to have a treatment ready for patients before the end of 2020.
“What our algorithms do is reduce the likelihood of drug-discovery failure in the lab,” said Satz. “We fail in the computer as much as possible to reduce the possibility of downstream failure in the laboratory. And that shaves a significant amount of time from laborious and time-consuming work.”
Some of the antibodies that the team is designing are intended to prevent coronavirus from attaching to the human body.
“The right-shaped antibodies bind to proteins that sit on the surface of human cells and the coronavirus, similar to a lock and key. Such binding can prevent the proliferation of the virus in the human body, potentially limiting the effects of the disease,” said Averso.
Additionally, by collaborating with the biotech industry, the pair is aiming to bring about therapeutics, diagnostics, and vaccines as quickly as possible.
As the threat of COVID-19 continues to surge throughout the US, other organizations are partnering to help combat the spread of the virus.
Recently, as part of a larger partnership with IBM, academic institutions, and national labs, Rensselaer Polytechnic Institute (RPI) offered researchers access to an artificial intelligence-powered supercomputer for COVID-19 research.
“In order to combat the devastating effects of this pandemic, we must be able to fully grasp the complexities and interconnectedness of biological systems and epidemiological data, as researchers work to develop therapeutic interventions and address gaps in our knowledge,” said Rensselaer President Shirley Ann Jackson.
“This effort requires expertise, collaboration, and the ability to process incredible amounts of data, and Rensselaer is offering all three at this critical time. In particular, the ability to model at very large scales requires the unique capabilities of AiMOS.”
With these machine learning algorithms, Satz and Averso expect to accelerate the speed at which coronavirus therapies are discovered, developed, and delivered.
“We are building a company that sits at the frontiers of AI and biotech,” Satz says. “We are hard at work accelerating the speed at which healing is discovered and delivered and could not ask for a more fulfilling mission.”