Alumni Spotlight: Bruce Weir

Bruce Weir earned a Ph.D. in statistics and genetics from NC State in 1968. During his tenure as a faculty member at NC State, he was a William Neal Reynolds Distinguished Professor of Statistics and Genetics and was founding director of the Bioinformatics Research Center. He also received the O. Max Gardner Award, the highest faculty honor presented by the Board of Governors of the University of North Carolina to recognize a faculty member having “made the greatest contribution to the welfare of the human race.”

Weir is one of the world’s foremost researchers on statistical analysis of DNA for forensic, human health and agricultural applications. He developed statistical tests of the frequency of genetic profiles that are now the standard for evaluating DNA evidence in forensic cases. He spearheaded reforms to the national guidelines on forensic DNA analyses, was instrumental in achieving acceptance of DNA evidence in courtrooms and is co-author of the definitive textbook on statistical inference in forensics. (Weir may be best known outside academic circles for testifying at the O.J. Simpson murder trial in 1995.)

He also established statistical measures of genetic linkage that are critical for mapping genes associated with human disease. The methods are widely used in gene mapping studies of complex traits such as diabetes and longevity. Weir’s seminal work in genetic descent and recombination in inbred and mixed populations laid the foundation for describing genetic differentiation among groups of animals and plants and has led to continuing improvements in crop and livestock breeding. After he moved to a faculty position at the University of Washington (UW), he brought with him the Summer Institute of Statistical Genetics, which he started at NC State. He served as the Chair of the Department of Biostatistics at UW from 2006 to 2014 and founded the UW department’s Genetic Analysis Center, which provides statistical expertise for large research consortia across the country. He is currently the director of the Institute of Public Health Genetics at UW, which offers graduate students exposure to faculty in medicine, law, ethics and other fields. He gave the esteemed 2016 C. Clark Cockerham Lecture at NC State.

What was it like to be a Ph.D. student of Clark Cockerham? You came to NC State from New Zealand to obtain a Ph.D. in Statistics and Genetics and study specifically under him. Would you say that you both had a good personality fit, or research fit, or both? Or what other aspects or practices allowed you to have a successful mentor-mentee relationship while you were a graduate student?

Working with Dr. Cockerham was an amazing experience. He had deep intuition about statistical genetics and he had the highest professional standards. He was completely focused on his research and he stressed quality. I was fortunate that his style was to work in a collaborative fashion with his students, and we spent many hours at the blackboard in his office. It was an exciting time.

What was it like transitioning from being a Ph.D. student of Clark Cockerham to his faculty colleague at NC State in terms of research collaboration and the mentor-mentee relationship? How did you differentiate your success as a researcher from your mentor’s success? How did you maintain such a productive and successful collaboration for 31 years, from the beginning of your Ph.D. program in 1965 until his death in 1996?

I took a postdoc position at UC-Davis after graduating and then returned to New Zealand as a faculty member. I didn’t return to NC State until 7 years after graduation, but during that time Clark and I continued to work together so when I did return it was already as a faculty colleague. I did need to establish some independence, though, and so I obtained my own R01 funding as well as continuing on the P01 that Clark directed. I had the best of both worlds.

While being a mathematics major at the University of Canterbury, you interned over the summer with Brian Hayman, a statistician who was studying genetic data, and that fateful opportunity led you to NC State to study with Clark Cockerham. What was it like being an international student at NC State at that time? How did you maintain your connections with your family back at home and how did you integrate in the department to make new friends and associates?

NC State and Raleigh in the period 1965-68 were certainly different than they are now, and travel to New Zealand (or even telephoning) was not as easy or affordable as now. The statistics department had several students from other British Commonwealth countries (U.K., Australia, South Africa, Canada) who formed a support community as I established U.S. friends and colleagues.

You were exposed to statistics and genetics in a summer internship and went on to become a world-renowned researcher on the statistical analysis of DNA for forensic, human health and agricultural applications. What do you think made you most successful as a researcher in statistical genetics? What is a guiding principle of your research process and the questions that you pursue?

The field of genetics has offered huge opportunities to people with mathematical and statistical interests. Both at NC State and now at UW I have been associated with talented people with similar interests. And I started with the experience of having worked with Clark.

You have been integral to the field of forensic science, where you brought your acumen and statistical and genetic knowledge to a field that lacked scientists with these background at the time. You have made major contributions including leading the effort to correct the erroneous sections in the first National Research Council (NRC) report on DNA methods in forensic science, 1991, and your views and methods were adopted in the second NRC report, 1996. What did it take to lead such an effort and to get the buy-in from the forensic science community and the National Research Council?

I did little besides forensic genetics in the 1990s. It was a period of great progress for typing technology development but some reluctance to accept statistical genetic arguments. I had some conflicts with statistical geneticists whose work I had greatly admired, but I believe the field has settled down considerably. The need for sound statistics remains.

You have served as an expert witness on the statistical reliability of DNA tests in several trials and pretrial hearings, including a trial in 1994 in Hillsboro, Ore., that resulted in a murder conviction, and the O.J. Simpson trial. You were in the public eye as an expert witness and were tasked with using your scholarly expertise to testify and write lengthy reports that are still publicly available today. What advice do you have for other statisticians who may be called as expert witnesses on trials, or involved in any other highly publicized endeavors?

I think we have an obligation to speak out for sound statistics. Our adversarial court system does expose us to criticisms at a level we don’t see in academia, but I believe we still have an obligation.

What are some ways that statisticians could improve on how they communicate statistical analyses to the public?

That is difficult, but I am impressed by my colleagues who have mastered the use of social media.

What are the current trends of the statistical analysis of DNA in forensic science?

The most interesting statistical development has been in probabilistic genotyping. Instead of declaring a particular genetic variant to be present or absent in an evidentiary profile, leading to inclusion or exclusion of a suspected contributor to the profile, we can now attach probabilities to the presence of a variant. The work rests on some sophisticated modeling and the use of MCMC analyses so there is the danger of interpretation appearing to be from a black box.

What are the research questions that you are interested in right now?

I am continuing work on a unified characterization of genetic population structure and relatedness that I described in my Cockerham Lecture. I’m trying to see how these new measures can be incorporated into linear mixed models for associating traits with genetic profiles. We are also looking at Bayesian implementations of our new measures. In the forensic work I’m looking at the interpretation of Y-chromosome profiles.

Where do you see the field of statistical genetics going in the next 10 to 20 years? What are topics that you think current students should be studying now to be competitive in this field in the future?

Statistical geneticists need to develop new tools for the interpretation of whole-genome sequence data and for integrating other -omic data (proteomic, metabolomics, etc). Students with an interest need a strong foundation in mathematics and statistics, skills in statistical computing and some understanding of molecular biology.

What’s the best advice that you can give to current graduate statistics students at NC State?

Take advantage of the wealth of talent among your faculty and your fellow students. Use your access to this talent to secure a broad training.

Any parting thoughts?

NC State Statistics is a jewel among statistics departments. I’m proud to be an alumnus.