METHOD AND COMPUTER-PROGRAM FOR CONTROLLING THE FAMILY-WISE ALPHA-RATE IN GENE ASSOCIATION STUDIES INVOLVING MULTIPLE PHENOTYPES

Multiple significance testing involving multiple phenotypes is not unc ommon in the context of gene association studies but has remained larg ely unaddressed. If no adjustment is made for the multiple tests condu cted, the type I error probability will exceed the nominal (per test) alpha level. Nevertheless, many investigators do not implement such ad justments. This may, in part, be because most available methods for ad justing the alpha rate either: 1) do not take the correlation structur e among the variables into account and, therefore, tend to be overly s tringent; or 2) do not allow statements to be made about specific vari ables but only about multivariate composites of variables. In this pap er we develop a simulation-based method and computer program that hold s the actual alpha rate to the nominal alpha rate but takes the correl ation structure into account. We show that this method is more powerfu l than several common alternative approaches and that this power advan tage increases as the number of variables and their intercorrelations increase. The method appears robust to marked non-normality and varian ce heterogeneity even with unequal numbers of subjects in each group. The fact that gene association studies with biallelic loci will have ( at most) three groups (i.e., AA, Aa, aa) implies by the closure princi ple that, after detection of a significant result for a specific varia ble, pairwise comparisons for that variable can be conducted without f urther adjustment of the alpha level. (C) 1998 Wiley-Liss, Inc.