TABLE 3.
Genotype and analyses characteristics of the 12 eligible publications.
| Array type | Array name | QC | Population stucture/PCA | Imputed | References Panel | Final SNPs number | Association | Covariates/Confounding variables adjusted for | Suggestive sig. | Replication | References |
|---|---|---|---|---|---|---|---|---|---|---|---|
| GW genotyping array | Affymetrix ® GW Human SNP Array 6.0 | Yes | Structure = Non parametric clustering of genotypes using the AWClust algorithm; PC = using eigenstrat method | No | NA | 8,08,465 | LoRM under an additive model for HTN (PLINK V1.04); LRM for SBP and DBP (PLINK V1.04) | Age, Sex, BMI, First 2 PCs of the genotypes | p < 1E-4 | Yes | Adeyemo et al. (2009) |
| GW genotyping array | Affymetrix GW Human SNP Array 6.0 array | Yes | Structure = MDS; PCs were calculated for all unrelated individuals and predicted forrelated individuals | Yes | Merged: YRI and CEU HapMap-phased haplotypes | 2,500,000 | LRM under an additive model | Age, Age2, Sex, BMI, First 10 PCs | p < 1E−6 | Yes | Fox et al. (2011) |
| GW genotyping array | Affymetrix GeneChip SNP Array 6.0 | Yes | PCA | Yes | Merged: HapMap Phase III and the Human Genome Diversity Project | ∼2,420,000 | LRM/LMM for family data under additive model | Age2, Sex, BMI, First 10 PCs | p < 1E−5 | Yes | Franceschini et al. (2013) |
| GW genotyping array | Affymetrix Axiom Biobank array designed specifically for the MVP | Yes | PCA using FlashPCA; Clusted by race using K-means approach K1-4 | Yes | 1000G phase 3, V5 | NR | LRM with additive models (untransformed), using SNPTEST-v2.5.4-beta | Age, Age2, Sex, BMI, First 10 PCs | p < 1E−6 | Yes | Giri et al. (2019) |
| GW genotyping array, GW sequencing | Illumina HumanOmni 2.5M BeadChip array(UGR); Illumina HumanOmni Multi-Ethnic GWAS/Exome Array (MEGA pre-commercial v1) using the Infinium Assay a.k.a MEGA Array (DDS and DCC and some of AADM); Affymetrix Axiom GW PanAFR Array Set (AADM) | Yes | PCA; fineSTRUCTURE analysis (Lawson et al., 2012) | Yes | Merged: Uganda Genome Resource, AGVP, 1000G phase 3 | >24,423,923 | LMM with two random effects | Age, Age2, Sex | Nonea | No | Gurdasani et al. (2019) |
| GW genotyping array | Affymetrix Axiom array | Yes | PCA; LMM using estimated kinship matrices with leave one chromosome out (LOCO) to account for population substructure and cryptic relatedness with Bolt-LMM | Yes | 1000G phase I | >2,696,785 | LMM using estimated kinship matrices in Bolt-LMM | Age, Age2, Sex, BMI, First 2 ancestry PCs | p ≤ 1E-2 | Yes | Hoffmann et al. (2017) |
| GW genotyping array | Affymetrix GW Human SNP Array 6.0 platform | Yes | Family structure modeled using LME models; PCA using EIGENSTRAT | Yes | Merged: HapMap phase 2 CEU and YRI data | ∼2,740,000 | LoRM under an additive genetic model In PLINK | First 10 PCs | p < 1E−6 | Yes | Lettre et al. (2011) |
| GW genotyping array | Either Affymetrix or Illumina | Yes | PCA | Yes | 1000G Phase 1 | >30,072,738 | LRM for unrelated data or by the generalized LMM for family data, under an additive genetic model | Age, Age2, Sex, BMI, up to first 10 PCs | p < 1E−6 | Yes | Liang et al. (2017) |
| GW genotyping array | Illumina HumanOmni2.5M-8 (‘octo’) Beadchip arrays, version 1.1 (Illumina Inc., San Diego, United States) | Yes | Structure = LMM (GCTA V1.22) | Yes | Merged: 1000G, AGVP and UG2G | 20,074,711 | LMM regression (accounting for population substructure) (GCTA V1.22) | Age, BMI | p < 1E−6 | Yes | Lule et al. (2019) |
| GW genotyping array | Infinium Multi‐Ethnic Global BeadChip (v1.0) | Yes | NA; TransMeta, follows a hierarchical framework that incorporates a kernel matrix K into the covariance structure of the effect size estimates | Yes | EA: HRC; AA, MA & JA: 1000G Phase 3 | 3,153,931 | LMM using GCTA (Log & rank‐based inverse normal transformations) | Age, Sex & self‐reported diabetes status (FEs) | p < 1E–6 | No | Willems et al. (2020) |
| GW genotyping array | MEGA array | Yes | PCA, (SNPRelete in R) | Yes | 1000G phase 3 | 28,263,875 | LoRm and LMM, in both SUGEN and GENESIS program | AHM adjustment, (for BP traits only, not HTN), First 10 PCs | Noneb | Yes | Wojcik et al. (2019) |
| GW genotyping array | Either with Affymetrix or the Illumina whole-genome SNP genotyping arrays | Yes | NA | Yes | NA | ∼2,420,000 | LRM, additive model; Plink | None | p < 5 × 10–7 | No | Zhu et al. (2015) |
GWAS sig. = p < 5 × 10–8 for all studies, except for aGWAS sig. = < 5E–9) bGWAS sig. = < 3E–9 for MAF <5%.