Award-winning research goes deep with neural networks for drug safety
Researchers Austin Daniels, Chris Calderon and Ted Randolph
A team from the Department of Chemical and Biological Engineering received one of the pharmaceutical industry’s top honors—the Ebert Prize—for using deep neural networks to analyze images of sensitive protein-based drugs. The Ebert Prize, established in 1873, is the oldest continuously awarded pharmacy award in the United States.
Lead author Adjunct Assistant Professor Chris Calderon, Gillespie Professor Ted Randolph and graduate student Austin Daniels authored the prize-winning research, in the April 2018 issue of the Journal of Pharmaceutical Science.
Protein-based drugs comprise the fastest-growing class of pharmaceuticals and provide remarkable benefits for human health, but they are also highly susceptible to damage while traveling from the production line to the patient.
“Different protein stresses—including mechanical agitation and temperature shock—can cause protein therapeutics to degrade,” Calderon said. “Certain types of stressed protein drugs can even cause adverse reactions, including death. Since protein therapeutics are often sensitive to their environment, characterizing their stability and quality at different stages of production and storage is crucial.”
The team used an innovative combination of deep neural network image analysis—utilizing machine-learning algorithms for image and speech recognition—with computational statistics, biophysics and protein engineering methods to evaluate and characterize impurities and contaminants within proteins engineered for pharmaceutical use.
“We demonstrate a new approach for extracting information about the quality and stability of protein-based drug formulations from images measured with a high-throughput, microfluidic microscope,” Calderon said.
“Facial recognition technologies and other machine learning techniques may seem far apart from traditionally used methodsĚýto develop stable formulations of protein drugs,” Randolph elaborated. “But the beauty of combining these seemingly disparate approaches is that it generates new insights into challenging problemsĚýand opens up whole new avenues for research. For example, we are now examining how the combined techniques may address issues associated with neonatal blood infections.”
Calderon credits the diversity and positive collaborative environment of the department for fostering this combined-discipline approach. Glenn L. Murphy Endowed Professor Daniel Schwartz connected Calderon and Randolph when he realized that the two professors might benefit from working together.
“The resulting collaboration has taught me a lot about protein therapeutics and continues to show promise in multiple avenues,” Calderon said. “Our collaboration probably wouldn’t have happened without Dan catalyzing this meeting, and it serves as a nice example of the interdisciplinary research potential at ChBE.”
The team will receive the Ebert Prize at the Annual Meeting of the American Pharmacists Association in Seattle in March.