Post-GWAS analysis approaches uncover dozens of novel genes associated with myopic refractive error

Abstract

Purpose : Our aim was to identify genes influencing refractive error, a very common eye disorder and an important cause of blindness. To complement ongoing (unpublished) standard per SNP (single nucleotide polymorphism) analysis, we conducted a range of post-GWAS (genome wide association study) approaches using data on myopia related traits in two large datasets.

Methods : We firstly considered refractive error (measured as quantitative trait spherical equivalent) in 44000 samples from the Consortium for Refractive Error And Myopia (CREAM); standard quantitative trait analysis was used. Secondly, we assessed myopia via questionnaire data from 23andMe; 104000 genotyped participants were asked "What age did you first wear glasses for myopia" and a survival analysis approach was used.

We applied three post-GWAS approaches. Firstly, we applied a gene-based test which accumulates evidence for association across all SNPs in/within 50kb of a gene (fastBAT). Secondly, we applied a gene-based test which incorporates eQTL information (Eugene). Genes were declared significant for the gene-based tests if they achieved P<2e-6 (correcting for number of genes). Thirdly, we applied an approach, where 450 ENCODE annotations are tested for enrichment in the single SNP GWAS (fgwas). In this third case, we also re-weighted our results based on the most important annotations to test for additional loci influencing myopia.

Results : Initially, the gene-based approaches were applied to the CREAM dataset. The two tests uncovered 5 genes not significant (P>5e-8) in standard per-SNP tests. We then validated these results using per SNP tests in the meta-analysed CREAM and 23andMe datasets. All of the novel gene-based associations replicated in per SNP tests (P<5-e8) for CREAM+23andMe.

Applying the gene-based approaches to the full CREAM+23andMe data revealed a total of 24 novel associations. The functional annotation approach identified several significantly enriched annotations (most significant was regions defined as being DNase I hypersensitive in fetal brain tissue); reweighting using these annotations uncovered a further 8 loci not significant in per-SNP tests.

Conclusions : Overall, our post-GWAS approach provided evidence for a role of a large number of novel genes in myopic refractive error.

This is an abstract that was submitted for the 2017 ARVO Annual Meeting, held in Baltimore, MD, May 7-11, 2017.