A comparison of power to detect a QTL in sib-pair data using multivariate phenotypes, mean phenotypes, and factor scores

The power to detect a quantitative trait locus (QTL) in sib-pair data is in vestigated. We assume that we have at our disposal 3 or 4 related phenotypi c measures in a sample of sib-pairs. Individual differences in these phenot ypes are due to a common QTL and specific (i.e., unique to each phenotype) nonshared environmental and specific polygenic additive effects. In additio n, models are considered that include common nonshared environmental effect s and/or common polygenic additive effects. We calculate the power to detec t the QTL in a genetic covariance structure analysis of the multivariate da ta, of the mean phenotypic data, and of factor scores. The use of factor sc ores is shown to be universally more powerful than the use of multivariate or mean phenotypic data. We also investigate the effect of using a single s ample of sib-pairs to both calculate the factor score regression matrix and to carry out the QTL analysis. The use of a single sample to both these en ds results in a loss of power compared to the theoretical, expected power. The gain in power attributable to the use of factor scores, however, outwei ghs this observed loss in power. The advantages of using factor scores in s electing sib-pairs are discussed.