The first assessment of cross array-induced imputation bias combined study subjects genotyped on 1M from SAGE (Bierut et al. 2010) with subjects genotyped on 550v3 from the Cancer Genetic Markers of Susceptibility (CGEMS) Pancreatic Cancer Cohort Consortium (PanScan) (Amundadottir et al. 2009) for European Americans or iControl for African Americans. Figure 1 presents the λgc values and percentages of SNPs with false positive associations (P < 1 × 10−6) for each of the three assessments of bias (genotyped SNPs shared on both arrays, imputation based on the union of genotyped SNPs across arrays, and imputation based on the intersection of genotyped SNPs across arrays). Tests of association between arbitrarily assigned case status and the genotyped SNPs shared on both arrays showed no statistically significant associations (Fig. 1a, d), indicating that there was no genotyping bias between these arrays. However, association tests for SNPs imputed based on the union of genotyped SNPs available on either array revealed spurious results as indicated by inflated λgc values and SNPs having statistically significant P values across the MAF spectrum (Fig. 1b, e). Overall, 0.20
