Duke Computer Science: Research, Startups and Event Horizons

June 18, 2020

By Glenn McDonald

For more than 40 years now, Duke Computer Science has been known as a place where people can get things done. The department has a celebrated tradition of generating projects, initiatives and startup companies that spin up and out from campus, bringing critical solutions into the larger world.

Ashwin Machanavajjhala: Tumult LabsDuke Computer Science associate professor Ashwin Machanavajjhala is piloting one of these startups just now. Tumult Labs is an industry leader in the busy arena of data management and privacy. The company’s core technology is designed to promote the free flow of information while protecting the privacy of all users.

Anyone who uses the Internet can tell you, it’s decidedly spooky how much online merchants know about us. Those custom ads online can be genuinely startling. How did they know I’m looking for a lawnmower? A red lawnmower!

This kind of retail profiling is just the tip of the iceberg, Machanavajjhala says. Information brokers are using superpowered machine-learning systems to tease out sensitive information from the colossal piles of aggregate data available online.

To get more specific would require a Ph.D. or three, but Machanavajjhala says that there is a solution. Tumult Labs’ specializes in a technique called differential privacy. The idea is to introduce a kind of randomized noise into data sets, which thwarts attempts at squirreling out sensitive information.

For several years now, Machanavajjhala has been working as a consultant with the United States Census Bureau, helping design methods to safely share statistics about people and establishments that enable policy making while ensuring privacy. Through Tumult Labs, Machanavajjhala aims to develop products that every government agency and business in the world can use to safely share data. In its first year, Tumult Labs is already deploying its privacy solutions at multiple government agencies.

“We are the world leaders in addressing this problem,” Machanavajjhala says.

Bruce Donald: Ten63Another startup initiative has recently emerged from the Donald Lab at Duke University, home to researchers from half a dozen different departments, including Computer Science, Biochemistry and Computational Biology. Ten63 Therapeutics is ramping up quickly, and the timing couldn’t be better.

Ten63 is dedicated to developing new and better therapeutics against the world’s most dangerous diseases. More specifically, the company deploys advanced computer modeling to target “undruggable” diseases for which there is no existing treatment. “At Ten63 we are leveraging our proprietary computational platform which allows us to search more chemistry, by orders of magnitude, than anyone else in the world,” says Marcel Frenkel, co-founder and CEO of Ten63 Therapeutics.

The foundation of Ten63’s proprietary models and algorithms were developed over a period of 15 years of NIH-funded research at the Donald Lab. The work has resulted in roughly 100 high-impact peer-reviewed publications, over 8,000 citations and designed molecules in 14 clinical trials.

"The Duke computer science department provided a de facto innovation hub to spin off this translational company into the therapeutics space,” says Prof. Bruce R. Donald, co-founder and CSO. ”All of the founders have doctorate degrees from my laboratory, and we have a shared vision of how our platform can radically advance computer-aided drug design and biopharmaceutical discovery.”

Across campus, at the BioLabs North Carolina building in downtown Durham, Duke professor Alvin Lebeck is keeping busy with his startup enterprise, Phitonex. Founded in the summer of 2017, Phitonex is developing fluorescent labels that increase the capabilities of flow cytometry, a process by which scientists can measure specific properties of cells. If it seems like such technology might be particularly important these days, well, Dr. Lebeck concurs.

Alving Lebeck: Phitonex“It's kind of a timely moment right now,” says Lebeck, professor of computer science and electrical engineering. “We're building what we call fluorescent labels, which are used to help researchers and scientists understand things about single cells.”

Flow cytometry is especially useful in studying the immune system, a critical element in understanding the human body’s response to the coronavirus. 

Lebeck says his success with Phitonex is a result of doing a very fundamental kind of research. “If I could get a message to the larger world out there, it’s about the importance of support for high-risk research with long term goals,” Lebeck says.

Such open-ended research, Lebeck says, can solve problems out on distant horizons that no one can see. “It’s like the space program in the 1960s, look at how many things came out of that,” Lebeck says. “There's a lot of value in supporting your local lab.”