Fig. 2
From: Challenges at the APOE locus: a robust quality control approach for accurate APOE genotyping
Limitations in APOE filtering approach 1 are reflected in discordance between imputed and provided APOE genotypes, particularly in APOE*4/4 carriers. A APOE*4/4-rs439401 carrier cohort distributions. The top section shows the distribution of prioritized APOE genotype source in approach 1, indicating that APOE*4/4 carriers of rs439401 had very few WGS/WES-verified APOE*4/4 data. The bottom section shows pie charts for carrier distributions across cohorts (Additional data in Table S8). The red arrow indicates that a large fraction of control rs439401 carriers was contributed by MIRAGE. B Concordance rates between provided and imputed APOE per cohort (additional data in Table S9). The red arrow indicates that MIRAGE had the lowest concordance rate, suggesting potential limitations with its provided APOE data that could explain observations in A. C Concordance rates between provided and imputed APOE for the discovery sample, considering multiple strata (additional data in Table S10). APOE*4/4 strata considered provided APOE*4/4 genotypes after applying APOE filtering approach 1. Note decreased concordance in APOE*4/4 controls compared to cases. Note strongly decreased concordance for rs439401 carriers, specifically controls. Simulations confirmed that APOE*4/4 controls are more likely than cases to not actually be APOE*4/4 carriers (cf. Fig. S6-7). Abbreviations: CN, cognitively normal; AD, Alzheimer’s disease; OR