Premium
Correlation of Population Parameters Leading to Power Differences In Association Studies with Population Stratification
Author(s) -
He Y.,
Jiang R.,
Fu W.,
Bergen A. W.,
Swan G. E.,
Jin L.
Publication year - 2008
Publication title -
annals of human genetics
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.537
H-Index - 77
eISSN - 1469-1809
pISSN - 0003-4800
DOI - 10.1111/j.1469-1809.2008.00465.x
Subject(s) - statistical power , population stratification , population , divergence (linguistics) , correlation , statistics , genetic association , genetic divergence , allele , biology , population size , population genetics , evolutionary biology , genetics , demography , mathematics , single nucleotide polymorphism , genotype , genetic diversity , linguistics , philosophy , geometry , sociology , gene
Summary The power of statistical tests to measure effect sizes in the presence of population stratification is an important issue for the design and analysis of population‐based association studies. Comparisons of statistical tests have shown that the power of different statistical approaches varies in different genetic scenarios. However, the impact of stratified population parameters on statistical power is not yet understood in a general statistical framework, particularly the impact of correlated population parameters. To investigate such impact in detail, we implemented a genetic model for population‐based association studies with stratified samples and evaluated the impact on power with different genetic scenarios. The investigation shows that correlation between disease prevalence and risk allele frequency among subpopulations impacts statistical power. In a model with five subpopulations and moderate population divergence ( Fst = 0.01), the correlation accounts for more than 85% of power difference. Our results also show that the estimation of genetic effect for candidate loci is biased by population divergence. Beneficial alleles could be wrongly characterized as risk alleles when prevalence differences and divergences of risk loci are large among subpopulations.