Genomewide association studies (GWAS) are perhaps the newest wave in human genetic research.
The approach involves rapidly scanning markers across complete sets of DNA — or genomes — of many people to find genetic variations associated with a particular disease. Once new genetic associations are identified, researchers can use the information to develop better strategies to detect, treat and prevent disease.
But investigators need to answer key questions to satisfy themselves and grant reviewers. How many subjects are needed? Should you study more single-nucleotide polymorphisms (SNPs) and fewer subjects or the reverse? For any design/analysis plan, what are the expected false positive and negative rates?
The Section on Statistical Genetics has created software that gives UAB investigators the power to answer those questions. The new GWA Test Driver enables researchers to test-drive combinations of prospective designs and analyses and produce reports containing publication-quality plots and tables of power analyses and sample-size calculations ready to include in manuscripts and grant applications. The GWA Test Driver is available online at gwatestdriver.ssg.uab.edu.
Researchers can investigate up to 1 million-plus markers, or SNPs, across the genome in a GWAS and that increases the probability of finding something by chance. But to detect statistical and clinical significance, they have to determine the sample size they need. The GWA Test Driver does that.
“This software will permit more investigators more easily, quickly and effectively to test if scenarios they can propose for GWAS are likely to be sufficiently powerful and worth doing,” says David Allison, Ph.D., director of the Section on Statistical Genetics. “In turn, when they decide a study is worthwhile, it aids researchers in justifying those studies effectively in grant proposals.”
Nita Limdi, Ph.D., Pharm.D., associate professor of neurology and epidemiology, already has had discussions with GWA Test Driver lead programmer Jelai Wang about the value of the software and its robustness at estimating sample sizes.
Limdi’s primary research focus is to reveal the effect of genetic and environmental factors on drug response, specifically those associated with toxicity and lack of efficacy. Currently her research efforts are focused on tailoring warfarin to reduce the risk of hemorrhagic complications.
Limdi received help from Allison, Wang and others to obtain her first major grants — a K23 Mentored Patient-Oriented Research Career Development Award, which she was awarded in 2003, and an R01 grant awarded in 2007.
“I was successful on my first try for both of those grants; a big reason is because of the help they provide,” Limdi says. “They do cutting-edge methodology research, like the GWA Test Driver, and that’s critical for our grant applications.”
For example, Limdi says when a research body reviews a grant application, they can see well thought-out power calculations and know that the investigator has an idea of the exact number of patients needed for their study.
That combine with numbers generated by materials built on their own campus makes a highly competitive package, Lindi says.
“For example: I want to find the genes that affect blood pressure, and I want to find people who need to be treated before they become hypertensive,” Limdi says. “You work with a statistician and an epidemiologist to figure out if the study is doable. You want to show you are estimating the effect you will find and the number of people you need. That is done using tools like the GWA Test Driver. And if that tool is developed on your campus, you’re showing the NIH the intellectual capacity is available on your campus that’s going to help you solve the problem.”
Making it work
The Section on Statistical Genetics developed the GWA Test Driver using a grant from the UAB Health Systems Foundation General Endowment Fund. The group submitted the proposal and secured funding three years ago.
The GWA Test Driver can aid researchers in several ways:
- Search among real and simulated datasets for those with similar characteristics to the study-in-planning and view a pre-calculated power-analysis report.
- Upload their own preliminary data for a custom power-analysis report.
- Perform an analytic power calculation
- Follow links to other genetic power software.
“I think the HSF-GEF and our researchers will be happy with this product,” says Wang, system programmer lead in biostatistics. “It will help investigators know how many samples they need to recruit. We’ve implemented several power calculations in the Test Driver, one of which we invented. We’ve also implemented a very simple analytic power calculation, and we link to other people’s power calculation software. It’s a very advanced, thorough product.”