Fig. 1: Analysis workflow from POAG and IOP GWAS to causal regulatory mechanisms, genes, pathways, and cell types.
a POAG and IOP associations genome-wide (known and modest associations) were tested for enrichment among expression and splicing quantitative trait loci (e/sQTLs) in GTEx tissues and retina compared to permuted null sets of variants matched on confounding factors, using QTLEnrich (one-sided). In cases where enrichment was found, the lower bound number of e/sQTLs in a given tissue, likely to be true trait associations was estimated using an empirically derived, true positive rate (\(({\pi }_{1})\) approach. b Putative causal genes were prioritized per known POAG and IOP genome-wide association study (GWAS) locus by applying two colocalization methods (eCAVIAR, enloc) to all e/sQTLs from 49 GTEx tissues and retina eQTLs that overlapped each locus, followed by two-sample Mendelian Randomization (MR). Overlap of the colocalizing GWAS loci and e/sQTLs with Hi-C (3D chromosome conformation capture), cis-regulatory element (CRE), and super-enhancer (SE) regions from human retina was utilized to further prioritize causal genes. The human and eye images were created with BioRender.com. c All target genes of significantly colocalizing e/sQTLs (e/sGenes) or cell type-specific genes per trait were tested for enrichment in signaling and metabolic pathways (Reactome, KEGG), gene ontologies, and mouse phenotype ontologies using GeneEnrich (one-sided). The POAG cross-ancestry GWAS meta-analysis Manhattan plot was generated using QMplot (https://github.com/ShujiaHuang/qmplot). d Significantly colocalizing e/sGenes were tested for enrichment in specific cell types in single-nucleus RNA-seq data of glaucoma-relevant eye tissues, using ECLIPSER (one-sided). Cell type-specific genes were defined with cell type fold-change>1.3 and FDR < 0.1 per tissue. Cell type-specificity significance per GWAS locus set for a given trait was assessed against a null distribution of loci associated with unrelated, non-ocular traits, using a Bayesian Fisher’s exact test. Genes mapped to GWAS loci with a cell type-specificity score above the 95th-percentile of null locus scores were proposed as contributing to the trait in the enriched cell type. e Cell type enrichment for the POAG and IOP GWAS was corroborated using two regression-based methods that assess cell type-specificity of trait associations considering all associations genome-wide: stratified-LD score regression (S-LDSC) and MAGMA.
