BACKGROUND: With the beginning of the era of genome-wide association studies methods to obtain 'in silico' genotypes have gained importance. In this context, an evaluation of genome-wide power levels of current marker panels and the power gain achievable with imputed genotypes are of high interest. METHODS: Power for single-marker analysis of imputed genotypes is evaluated via a simulation study based on HapMap data. Power values for genome-wide significance of marker panels of 1,000,000 SNPs are considered for small effect sizes typical of common diseases and large case-control samples. In order to evaluate the performance of imputing, we consider a method that is conceptually related to previous approaches. We introduce various modifications which together lead to an alternative implementation of the imputation idea. In particular, a Monte-Carlo (MC) simulation method for association testing of imputed markers is introduced. RESULTS: We show that the incorporation of imputed genotypes can lead to a substantial power gain for common disease variants if the training sample is large enough. In addition, we show that the MC approach is valuable to for validating association results obtained with imputed genotypes. DISCUSSION: Our simulation study also shows that even denser marker panels than those currently available are needed when sample size is limited. We thus expect that full genome SNP panels will lead to the identification of additional disease variants in the future. Until then, it is desirable that large and ethnically matched training samples genotyped on dense marker panels are available in each country. (c) 2009 S. Karger AG, Basel.
BACKGROUND: With the beginning of the era of genome-wide association studies methods to obtain 'in silico' genotypes have gained importance. In this context, an evaluation of genome-wide power levels of current marker panels and the power gain achievable with imputed genotypes are of high interest. METHODS: Power for single-marker analysis of imputed genotypes is evaluated via a simulation study based on HapMap data. Power values for genome-wide significance of marker panels of 1,000,000 SNPs are considered for small effect sizes typical of common diseases and large case-control samples. In order to evaluate the performance of imputing, we consider a method that is conceptually related to previous approaches. We introduce various modifications which together lead to an alternative implementation of the imputation idea. In particular, a Monte-Carlo (MC) simulation method for association testing of imputed markers is introduced. RESULTS: We show that the incorporation of imputed genotypes can lead to a substantial power gain for common disease variants if the training sample is large enough. In addition, we show that the MC approach is valuable to for validating association results obtained with imputed genotypes. DISCUSSION: Our simulation study also shows that even denser marker panels than those currently available are needed when sample size is limited. We thus expect that full genome SNP panels will lead to the identification of additional disease variants in the future. Until then, it is desirable that large and ethnically matched training samples genotyped on dense marker panels are available in each country. (c) 2009 S. Karger AG, Basel.