Ben Koby stands in front of an out-of-focus whiteboard with chemistry formulas on it and smiles at the camera.

Koby Earns Biophysical Society Award, NIH Fellowship

By Kirsten Heuring Email Kirsten Heuring

Heidi Opdyke

Ben Koby computes new ways to develop potential drug candidate molecules. When the Critical Assessment of Computational Hit-Finding Experiments (CACHE) Drug Discovery Challenge #1 was announced, he was excited to participate.

Teams were tasked with using computational methods to find hits, potential drug targets, for leucine-rich repeat kinase 2 (LRRK2), a donut-shaped protein implicated in familial Parkinson's disease. Because of its shape, researchers face difficulties finding molecules that bind to LRRK2.

"The goal of the competition was to find novel, small molecules that bind to an orphan drug target," said Koby, a Ph.D. student in Carnegie Mellon University's Department of Chemistry. "The protein itself is the most common genetic cause of familial Parkinson's disease."

Koby, chemistry graduate student Filipp Gusev and recent Ph.D. alumnus Evgeny Gutkin were part of the Carnegie Mellon team formed and led by the Chemistry Professor Maria Kurnikova and Olexandr Isayev, Carl and Amy Jones Professor in Interdisciplinary Science.

The team developed an active learning cycle workflow with automatic machine Learning based on absolute free binding energy simulations to predict how molecules would interact with LRRK2. They then performed the simulations on the Pittsburgh Supercomputing Center's Bridges-2 supercomputer. Out of 800 molecules simulated, they chose seventy-six for experimental validation and found five that bind to LRRK2. In the second round of the competition, the team identified another eight experimentally active compounds based on their previous results.

CACHE was created to find the most efficient computational methods for initial hit-finding. In the first challenge, the Carnegie Mellon team tied for first place. Kurnikova said Koby has continued to make advances on the computational tools he helped develop for the challenge.

"Ben has made multiple improvements to various aspects of the workflow in the past two years," Kurnikova said. "For example, he developed an algorithm to streamline the free energy simulations, which are the bottleneck of the approach, to decide on the fly when to stop the simulation."

Koby presented his work at the Biophysical Society's Annual Meeting in February. He earned a Student Research Achievement Award for his outstanding presentation during the poster competition.

"Typically, poster sessions last two hours so that presenters can look at other posters. I was presenting for the full two hours," Koby said. "As a scientist, it always feels good to have recognition for what you've done."

Koby's work was also recognized by the National Institutes of Health (NIH). He earned the Ruth L. Kirschstein Predoctoral Individual National Research Service Award Fellowship, which funds doctoral students' dissertations and provides mentored research training. Koby will use the award to develop free energy-based multi-objective active learning machine learning methods. These methods will optimize compounds to make them bind more effectively for a single target or multiple targets.

Kurnikova said that Koby's NIH Fellowship was a major milestone that will support him as he continues his Ph.D. work.

"Ben is enthusiastic, meticulous and hard working. He loves to learn new things, and he is not deterred by setbacks," Kurnikova said. "In our line of work, one has to endure multiple failed attempts to improve on our models and understanding. Ben sees the light at the end of the tunnel and goes through with no hesitation."

Currently, Koby is participating in CACHE Drug Discovery #5 along with other members of the Kurnikova and Isayev groups. The team is working to investigate the structure of a G-protein coupled receptor, MCHR1, based on the compounds that bind to it. MCHR1 is implicated in obesity, sleep disorders, depression and anxiety. Koby said that he is excited about the challenge.

"This is going to be our first time doing these relative binding free energy calculations on a membrane protein," Koby said. "The competition aspect is really nice because it gives you a reason to wake up and get on it."