Imputation across genotyping arrays for genome-wide association studies: assessment of bias and a correction strategy

Abstract

A great promise of publicly sharing genome-wide association data is the potential to create composite sets of controls. However, studies often use different genotyping arrays, and imputation to a common set of SNPs has shown substantial bias: a problem which has no broadly applicable solution. Based on the idea that using differing genotyped SNP sets as inputs creates differential imputation errors and thus bias in the composite set of controls, we examined the degree to which each of the following occurs: (1) imputation based on the union of genotyped SNPs (i.e., SNPs available on one or more arrays) results in bias, as evidenced by spurious associations (type 1 error) between imputed genotypes and arbitrarily assigned case/control status; (2) imputation based on the intersection of geno-typed SNPs (i.e., SNPs available on all arrays) does not evidence such bias; and (3) imputation quality varies by the size of the intersection of genotyped SNP sets. Imputations were conducted in European Americans and African Americans with reference to HapMap phase II and III data. Imputation based on the union of genotyped SNPs across the Illumina 1M and 550v3 arrays showed spurious associations for 0.2 % of SNPs: ~2,000 false positives per million SNPs imputed. Biases remained problematic for very similar arrays (550v1 vs. 550v3) and were substantial for dissimilar arrays (Illumina 1M vs. Affymetrix 6.0). In all instances, imputing based on the intersection of genotyped SNPs (as few as 30 % of the total SNPs genotyped) eliminated such bias while still achieving good imputation quality.

A meta-analysis of genome-wide association studies identifies ORM1 as a novel gene controlling thrombin generation potential

“…41,45,46 In order to increase the sensitivity of our discovery phase, we also considered of potential interest any SNP that did not reach genome-wide significance but were nevertheless associated at P , 10 25 with at least 2 TGP biomarkers.…”

“…40 All SNPs with acceptable imputation quality (r 2 . 0.3) 29,41 and MAF . 0.01 in both imputed GWAS datasets were kept for association analysis.…”

Two Adjustment Strategies for Imputation across Genotyping Arrays

“…However, both strategies have their weaknesses. The union strategy results in spurious associations in the absence of a genotyping bias and is not a valid approach [1] . The intersection strategy does not introduce spurious associations [1] , but it results in a reduced statistical power due to leaving out some of the available genotyping data.…”

“…The correction strategy combines measured genotypes as well as imputed and corrected genotype dosages for SNPs available curacy compared to arrays with sparser coverage. Several studies have shown that imputation errors may bring about biases and spurious associations, and filtering SNPs based on their imputation quality or accuracy before performing association tests cannot eliminate all the false positives [1][2][3] .…”

“…However, if we simply combine the genotypes from different arrays, spurious associations due to imputation are likely to occur. Two strategies of imputation across genotyping arrays are possible: (1) imputation based on the union of genotyped SNPs (i.e., SNPs available on at least one of all arrays) [2] , and (2) imputation based on the intersection of genotyped SNPs (i.e., SNPs available on all arrays) [1] . However, both strategies have their weaknesses.…”