Table 2 Contributions toward phenotype imputation and empirical kinship application from the GWAS group
From: Methods and results from the genome-wide association group at GAW20
Contribution | Genotypes | Phenotypes | Evaluation | Quality Control |
|---|---|---|---|---|
Chen et al. [11] | Restricted simulated SNP genotypes. 1. Null scenario: 19,763 SNPs on chromosomes 21 and 22 2. Alternative scenario: 5 known causal SNPs (rs9661059, rs736004, rs1012116, rs10828412 and rs4399565) | Simulated TG levels (a) Average difference between pretreatment (visits 1 and 2) and posttreatment (visits 3 and 4) or (b) single difference between visits 1 and 3 of log-transformed TG levels | 1. Type I error rate evaluation in “null scenario” 2. Power evaluation in “alternative scenario” | No quality control (QC) conducted on restricted simulated data |
Blackburn et al. [12] | Genome-wide (autosome) SNP data from 822 subjects in 173 pedigrees | TG and HDL-C levels were averaged for pre-treatment (visits 1 and 2) and post-treatment (visits 3 and 4) and regressed on age, sex, their interactions (age × sex, age2, age2 × sex), study center, smoking, and principal components 1–4; resulting residuals were inverse normalized | Under 3 different kinship models: 1. Heritability analyses 2. Single variant association testing | Exclusion of 6 individuals with unexpected relationships Variants were uplifted to hg19 mapping coordinates, excluding 135 conversion failures |
Porto et al. [14] | Genome-wide (autosome) SNP data from 822 subjects in 173 pedigrees | Averaged TG levels of pre- (visits 1 and 2) and post-treatment (visits 3 and 4) | Genomic best linear unbiased prediction (G-BLUP) under 3 different kinship models | Exclusion of 6 individuals with unexpected relationships; variants were uplifted to hg19 mapping coordinates, excluding 135 conversion failures |
Peralta et al. [13] | Genome-wide (autosome) SNP data from 822 subjects in 173 pedigrees | Averaged and log-transformed TG levels pre-treatment (visits 1 and 2) and post-treatment and corresponding empirical genetic values (EGVs) from Porto et al. [14] Simulated traits with zero mean, unit variance and a 35% heritability, but not linked to any real loci | Multipoint variance component linkage analyses under the pedigree-based kinship model | Exclusion of 2 individuals with unexpected relationships and 1 monozygotic twin to guard against the artificial inflation of heritability estimates. Variants were uplifted to hg19 mapping coordinates, excluding 135 conversion failures; LD based pruning of r2 ≥ 0.9 and exclusion of variants of minor allele count (MAC) > 5 left 375,632 variants for analysis |