Dean Billheimer: An Ambassador for Statistics in the World of Bioscience

In the new biology, it’s not about the mountains of numbers, says Dean Billheimer, it’s whether they mean exactly what you think they mean.

In his first months as BIO5’s new statistics mentor, Billheimer has sharpened the statistical designs on more than 20 grant requests from across the campus.

“That’s my job,” says Billheimer, an associate professor of biometry in the College of Agriculture and Life Sciences and director of the BIO5 Statistics Consulting Laboratory, “to make science better.”

Billheimer, an expert in statistics applied to proteomics, has collaborated on some of bioscience’s biggest research endeavors, including a massive search to understand the elusive causes of childhood asthma and a complex study of protein signatures’ role in diseases.

“Statistics is a natural extension of science,” he said. “If you collect a lot of data, you are always going to have ambiguous results. What do you do next? How do you evaluate measurement errors? That’s where the consulting lab can help.”

Take the case of the protein signatures. When a research team measures the amount of certain proteins in each person in a large study, they look for molecular signatures that correlate with a specific disease or response to treatment. Such findings could aid early detection of a disease and spot whether a treatment was effective.

But few results can be more ambiguous than those in proteomics, a field where there are no easy standard units for proteins, like centimeters and kilograms. “To take information from the new measurement systems, you have to adapt the formulas,” Billheimer said. “You may think you are measuring a certain property and assigning meaningful numbers, but are you really?”

Since the 2002 Nobel Prize for chemistry was awarded for work that opened the path to the emergence of proteomics, Billheimer has been helping scientists sort out the measurement puzzles in an often baffling field. He has become an expert in the use of matrix-assisted laser desorption/ionization, or MALDI, a mass spectrometry technique often used to analyze large biomolecules like proteins and peptides.

Most of his partnerships at the UA are interdisciplinary, reflecting the state of the new biology’s frontiers. “There are questions in patient selection, in forming a meaningful sample, in how to draw a comparison group,” Billheimer said. “And when the equipment used is very costly and slow, the question of size of the sample is crucial. People have to have foresight.”

Unlike most statisticians, who enter the field through mathematics, Billheimer came to statistics by way of science and engineering. He worked at Pfizer, the pharmaceutical giant, making antibiotics. “I was awake at night figuring out if Fermenter Tank 702 had an abnormal problem,” he said. When he began a graduate engineering degree at New Mexico State University, he discovered statistical tools he says he could have applied at Pfizer. “I was like a sponge,” he recalled. “I knew the real world problems and these tools offered the solutions.”

He shifted to experimental statistics for his master’s degree and received his doctorate at the University of Washington, working on statistics for biological monitoring.

Later, he applied statistics to airplane safety and flight operations at the Boeing Company in Seattle. There he came up with a way to perform statistics with objects of unusual shapes.

As he spoke, Billheimer held a piece of paper flat, letting it droop at the center, then moved his fingers to the center as the edges drooped. “How do you do statistical quality control on airplane parts whose shapes change based on how you support them?” he asked. “Like a wing skin.”  To solve that problem at Boeing, he invented new statistics for a “gravity neutral shape.”

He went on to work in biostatistics and proteomics at the University of Washington, at the Vanderbilt-Ingram Cancer Center in Nashville, and at the University of Utah’s Huntsman Cancer Institute, where he was director of the biostatistics core facility.

He joined BIO5 in 2008, assigned to begin building a team for the Statistics Consulting Laboratory. “We will offer help to every department, all across the campus,” Billheimer said. “It’s a huge need.”

Collaboration Snapshot
Billheimer has worked with Serrine Lau, also a BIO5 member and director of the Southwest Environmental Health Sciences Center at the College of Pharmacy, on proteomics approaches to the study of biomarkers. Two of their projects are trying to identify biomarkers for diabetes and chronic kidney disease from plasma. That work looks for markers for preclinical disease, to spot high-risk patients and describe variables that can point to a certain course of treatment. “One question is how to make efficient use of experimental material, samples, and instrument time,” Billheimer said.

He has worked with Fernando Martinez, the interim director of BIO5, and with BIO5 researchers Steve Rounsley and Rod Wing on identifying single genomic variants, or mutations, that are associated with developing childhood asthma. That work often involves measurements of single nucleotide polymorphisms – SNP’s (or “snips”) – in DNA that account for genetic differences among individuals. He applies statistical tests to learn what polymorphisms are associated with diseases. The goal is to more efficiently identify genomic variants associated with disease, minimizing the number of false positives.

Accomplishments
Billheimer worked in industry for six years, as a process engineer at Pfizer and as an aerospace statistician at Boeing. At the UA, he works closely with scientists from many disciplines and has advised faculty on more than 20 grants, seven of them National Institutes of Health Challenge Grants.

At the University of Washington, he was a research scientist at the National Research Center for Statistics and the Environment. At Vanderbilt-Ingram Cancer Center, he was director of the High Dimensional Data Analysis (Proteomics) Core and worked with mass spectrometry proteomics profiling at Vanderbilt University School of Medicine. He was director of the Biostatistics Shared Resource at the Huntsman Cancer Institute in Utah.