A company started by Duke researchers harnesses advanced software to fight drug resistance with the goal of creating more effective cancer treatments.
Researchers from the lab of Duke's Bruce Donald, James B. Duke professor of computer science, launched Gavilán Biodesign to harness new software that allows them to evaluate trillions of molecules at once. IndieBio, a biotechnology startup accelerator, selected the company as one of its 11 startups.
The company’s co-founders include Donald and three students with Duke ties—Mark Hallen, Trinity '09 and Graduate School '16, Jonathan Jou, Trinity '09 and Graduate School '18, and doctoral candidate Marcel Frenkel.
“Our software is a combination of computational chemistry and artificial intelligence methods," Hallen said. "It is for drug discovery. We are hoping to basically look at a large slate of drug candidates that a pharmaceutical company might be considering and then find ones that have desirable properties that can avoid drug resistance.”
Donald explained that drug resistance can quickly emerge in cancer molecules. The researchers are "interested in finding techniques to predict resistance ahead of time" in the computational phase of their analysis, he added.
"This means more than just looking at what the tumor did previously," Donald said. "New chemistry will force the organism to evolve new escape mutations to avoid binding of inhibitors, and we want to be able to predict this ahead of time.”
However, the drug resistance-detecting technology did not originate in Gavilán Biodesign. The research originated in Donald's lab at Duke.
Donald elaborated that the researchers decided to take this software to the commercial stage because companies have the resources to become more connected to biopharmaceuticals and drug development. It is a challenge to do this in a university setting because software engineering is not valued as much as new ideas and algorithms, Donald said.
One of the biggest obstacles the lab has faced, he explained, is the level of validation needed to prove that their software will work.
When the software was provided to a cancer research group at University College London, it confirmed the group's resistance predictions about drugs that were going into clinical trials. The researchers said they take pride in the unique efficiency of their software.
“There are all sorts of fancy techniques in molecular modeling, like molecular dynamics, that study drug evolution," Donald said. "But our technique...gives extremely accurate quantitative predictions of resistance.”
Get The Chronicle straight to your inbox
Signup for our editorially curated, weekly newsletter. Cancel at any time.