Quantifying the percent increase in minimum sample size for SNP genotyping errors in genetic model-based association studies.

摘要: Kang et al. [Genet Epidemiol 2004;26:132–141] addressed the question of which genotype misclassification errors are most costly, in terms minimum percentage increase sample size necessary (%MSSN) to maintain constant asymptotic power and significance level, when performing case/control studies genetic association a model-free setting. They answered for single nucleotide polymorphisms (SNPs) using 2 × 3 χ2 test independence. We address same here model-based framework. The model parameters considered are: disease (dominant, recessive), genotypic relative risk, SNP (marker) allele frequency, linkage disequilibrium. %MSSN coefficients each six possible error rates determined by expanding non-centrality parameter distribution under specified alternative hypothesis approximate linear Taylor series rates. In this work we assume misclassifying one homozygote as another 0, since these thought rarely occur practice. Our findings that there settings lead large total both dominant recessive models. As minor approaches increases without bound, independent other parameters. general, is complex function Use SNPs with small frequency requires careful attention genotyping insure specifications met. Software perform calculations study design available, an example its use given.