Replication of past candidate loci for common diseases and phenotypes in 100 genome-wide association studies
Résumé
Genome-wide association studies (GWAS) have created a paradigm shift in discovering genetic associations for common diseases and phenotypes, but it is unclear whether the thousands of candidate genetic association studies performed in the pre-GWAS era had found any reliable associations for common diseases and phenotypes. We aimed to systematically evaluate whether loci proposed to harbor candidate associations before the advent of GWAS are replicated in GWAS. GWAS published through 8/2008 and included in the NHGRI catalog were screened and variants in candidate loci were selected on the basis of statistical significance (p<0.05) to create a list of independent, non-redundant associations. Altogether 159 articles on GWAS were evaluated, 100 of which addressed past proposed candidate loci. A total of 291 independent, nominally significant (p<0.05) candidate gene associations were assembled after keeping only the lowest p-value SNP for each locus and each phenotype; 108 of those had p<10-3 for association and 41 had p<10-7. Twenty-two of these 41 candidate gene associations pertained to binary phenotypes with a median odds ratio 2.91 (IQR 1.82-4.6) and median minor allele frequency 0.17 (IQR 0.12-0.29) in Caucasians; for comparison, 60 new associations of binary outcomes with p<10-7 discovered in the same GWAS had much smaller effects (median odds ratio 1.30, IQR 1.18-1.58) and modestly larger minor allele frequencies (median 0.27, IQR 0.15-0.43). Overall, few of the numerous genetic associations proposed in the candidate gene era have been replicated in GWAS, but those that have been conclusively replicated have large genetic effects that should not be discarded.
Origine : Fichiers produits par l'(les) auteur(s)
Loading...