Mighty math: Valedictorian Pogrebniak advances medicine through computation
By her own account, when Katherine Pogrebniak first arrived at the National Eye Institute in the summer before her sophomore year at Princeton, she knew relatively little about the highly complex research in the national laboratory.
By the end of the summer, she had sketched out the framework of a model to analyze a protein that researchers believe could be a critical factor in the development of an eye disease affecting children.
"She has this tremendous ability to learn," said Yuri Sergeev, a senior researcher at the Bethesda, Maryland, institute, which is a branch of the National Institutes of Health (NIH). He said Pogrebniak brought an intense focus to her work, first learning biological aspects of the project and then bringing her computational skills to bear on working out a solution.
"She is a very talented, very focused and very smart young lady," Sergeev said. "She also works very hard."
Pogrebniak's peers and professors speak of the intellect and determination that have led her to the top of her class. A computer science major from Jacksonville, Florida, she is the valedictorian of the Princeton Class of 2014, and will deliver an address at the University's Commencement ceremony on June 3.
"She is very intelligent and she also has a superb work ethic," said Mona Singh, a professor of computer science and the Lewis-Sigler Institute for Integrative Genomics. Singh, who advised Pogrebniak for her independent project, said her contribution to her lab was akin to that from an advanced graduate student.
In a second internship at the NIH last summer, Pogrebniak helped a team of researchers develop a new method of using magnetic resonance imaging (MRI) scans in the treatment of patients with multiple sclerosis (MS). Daniel Reich, a principal investigator at NIH and the team leader, called Pogrebniak "an outstanding young star," and said that through her work the lab was able to achieve a major goal in imaging research.
Earlier this month, the results of Pogrebniak's work at the NIH were presented at the annual meeting of the International Society for Magnetic Resonance in Medicine in Milan.
"Katherine made quite an impression on me and all my colleagues," Reich said in a letter to the University. He noted that in applying computation to solve biological problems, "none of us has any doubt that Katherine will become a world leader."
Math at an early age
Pam Crawford, chair of the mathematics department at Jacksonville University, remembers when she first heard of Pogrebniak.
Pogrebniak's father, who is a pediatric ophthalmologist, asked whether it would be possible for Pogrebniak to begin studying mathematics at the university, Crawford said. "She started in college algebra, which is the lowest level that we teach."
Pogrebniak was 8 years old, in the third grade.
"Our other students thought it was neat to have her in class; she made some good friends," Crawford said. "We were all pleased when she would come in after the weekend and say she had been to a sleepover."
Over the next few years, Pogrebniak took math at Jacksonville University and the rest of her classes in local schools.
"She just kept going," Crawford said. "By the end, she was taking graduate-level courses."
Pogrebniak started at Princeton when she was 16. Ever since, whenever she is home on a break, Pogrebniak stops in to see Crawford.
"She really encouraged me," Pogrebniak said. "She has been an inspiration."
In 2010, Pogrebniak and her father, Alexander, collaborated with Crawford on a research article in the journal Investigative Ophthalmology and Visual Science. The article used a statistical analysis to examine methods to predict a glaucoma-like illness in children.
Understanding the big picture
Pogrebniak is the first woman earning a bachelor of science in engineering degree at Princeton to be named valedictorian. She also will receive a certificate in engineering biology. Pogrebniak is a two-time winner of the Shapiro Prize for Academic Excellence, a recipient of the Princeton Accenture Prize for Computer Science and co-winner with salutatorian Alexander Iriza of the Class of 1939 Scholar Award, which is given to the student with the highest academic standing at the end of their junior year. She is a member of Tau Beta Pi, the engineering honor society, and Phi Beta Kappa. At Jacksonville University, she was inducted into Pi Mu Epsilon, the mathematics honor society.
For her independent research, Pogrebniak joined a group in Singh's lab, which is studying how proteins bind with other molecules, called ligands. These binding interactions are a fundamental process of biology. Understanding the mechanism that drives them could broaden knowledge of protein interactions and lead to new drugs and medical therapies.
"Proteins are the workhorses of the cell," said Dario Ghersi, a postdoctoral researcher who worked closely with Pogrebniak. "They are involved in most of the functions crucial to life."
Singh's team has been working to develop a computational method that could predict the hydrophobicity (the tendency to be repelled by water molecules) of a ligand given the composition of a protein-binding site. The team developed a massive data set: tens of thousands of binding sites, with each site containing multiple components, all of which exhibit varying degrees of hydrophobicity. The ligands also vary in hydrophobicity, adding even more complexity.
Taming massive data sets is one of Pogrebniak's talents. Singh assigned her to teach the computer to look for repeated measurements in the data that would skew the analysis by suggesting that certain interactions are more common than they really are.
"Her task was to essentially prune the data set," Singh said. "Computationally, it was very challenging."
Ghersi and Pogrebniak tried a variety of different strategies, and Ghersi said the solution was always going to be a tradeoff: eliminating every redundancy would take the computer too long, but missing too many could affect the final analysis.
"It's a trial-and-error process and you need to be clever," he said. "It is a compromise between being extremely accurate and fast enough that you can actually do the work."
Ghersi said the method works and the project is moving forward.
"I was very impressed with Katherine's degree of scientific maturity in dealing with the problems and coming up with alternatives," he said. "We tried many things to reduce the redundancy in this data set. Her persistence was key."
Computation and health
Pogrebniak speaks quickly and enthusiastically as she begins to explain her research and her passion to apply her skills to improving people's health. Colleagues uniformly speak of her consideration for fellow students and co-workers. She has served for two years as a resident adviser at Wilson College; Eduardo Cadava, an English professor and the college master, spoke of her "wild intelligence — which she wears with grace and humility."
Singh referred to her modesty, adding that "you would never know how much she has accomplished."
Pogrebniak said she has long wanted to work in a medical field and help patients both through research and clinically.
"I started with math," Pogrebniak said. "But I have always been interested in biomedical problems because I was excited about their real-world applications."
Pogrebniak will attend the University of Cambridge as a Churchill Scholar in the fall and plans to obtain a master's degree in computational biology. After that, she intends to pursue a medical degree and a doctorate at Stanford University.
Last summer, when she worked at the NIH's National Institute of Neurological Disorders and Stroke, she assisted with ways to formulate new treatments for people with MS. Reich, the lead researcher, said the team has been working on ways to use magnetic resonance imaging to measure the insulation around patients' nerves in the outer area of the brain. The insulation, called myelin, is critical in nerve function, and damage to the myelin is involved in a number of neurological disorders, including MS.
Researchers believe that damage to myelin in the outer layer of the brain, called the cortex, plays a critical role in the development of MS. But myelin's relative scarcity in the cortex, combined with the cortex's complex folds, has made studying degradation of myelin in the cortex extremely difficult.
Recently, researchers have developed ways to use MRI scans to detect myelin in the cortex, and Reich's team wanted to use the scans to find areas with damaged myelin in patients. The software tools used to analyze the imaging were not designed for this, but Pogrebniak was able to find ways to adapt them. In fact, Reich said in his letter to Princeton, she came up with a method to use errors in the scan data to identify lesions in the cortex — "a major but hitherto unachieved goal of contemporary MS-imaging research."
"In imaging, there is a lot of physics and math, but most of the people who do imaging can't deal with massive amounts of data," he said. "Katherine is certainly someone who can do that."