2DkcTXceO

N. Chatterjee 183by any means. I just stumbled upon the problem when I was attempting to dobasic power calculations for a collaborator. Looking back, it was a risky thingto undertake as I was not sure where my effort was going to lead to, otherthan maybe I could advise my collaborators a little more intelligently aboutstudy designs. It was more tempting to focus, like many of my peers have done,sometimes very successfully, on so-called “hot” problems such as developing anoptimal association test. Although I have put some effort in those areas as well,today I am really glad that instead of chasing more obvious problems, I gavemyself the freedom to venture into unknown territories. The experimentationhas certainly helped me, and hopefully the larger scientific community, toobtain some fresh insights into study designs, statistical power calculationsand risk-prediction in the context of modern high-throughput studies.17.5 The post-GWAS era: What does it all mean?It is quite amazing that even more than two decades after the BRCA1/2mutations were discovered, there is so much ambiguity about what are the truerisks associated with these genes for various cancers. In the literature, availableestimates of lifetime-risk of breast cancer, for example, vary from 20–90%. Asnoted above, while estimates available from highly-enriched cancer familiestend to reside at the higher range, their counterparts from population-basedstudies tend to be more at the lower range. Risk for an individual carrier wouldalso depend on other information, such as the specific mutation type, cancerhistory among family members and information on other risk-factors. Theproblem of estimation of risk associated with rare highly penetrant mutationsposes many interesting statistical challenges and has generated a large volumeof literature.Discoveries from genome-wide association studies are now fueling the debatehow discovery of low penetrant common variants can be useful for publichealth. Some researchers argue that common variants, irrespective of howmodest their effects are, can individually or collectively highlight interestingbiologic pathways that are involved in the pathogenesis of a disease and hencepotentially be useful for development of drug targets. Although this would bea highly desirable outcome, skepticism exists given that discoveries of evenmajor genes like BRCA1/2 have seldom led to successful development of drugtargets. Utility of common variants for genetic risk prediction is also now amatter of great debate. While a number of early studies painted mostly anegative picture, large numbers of discoveries from the most recent very largeGWAS suggests that there is indeed potential for common variants to improverisk-prediction.I am an avid follower of this debate. While the focus of the genetics communityis quickly shifting towards what additional discoveries are possible using