Finally, a central issue in the field of genetic association studies is the lack of consensus on how to deal with multiple, non-independent analyses. Because there is likely to be LD between many of the individual markers examined, a standard Bonferroni correction (which assumes complete independence) could mask the existence of some, if not many important associations. As such, statistical geneticists have developed a number of alternative strategies for taking these dependencies into consideration (see Ziegler et al., 2008 for a recent review). One fairly common approach is to use the existing LD structure to estimate the number of “independent” effects represented by a set of SNPs (Nyholt, 2004). That is, the estimated number of effects is used as the denominator in a modified Bonferroni correction. A software package for this purpose, SNPSpD, is freely available on-line4. Still, since there is no agreed upon gold standard for dealing with multiple testing, it is probably most important for researchers to present sufficient information for a diverse audience to be able to assess the approach taken, and the interpretation proffered.
