Fig. 2: Gene prioritisation for non-MHC pneumonia susceptibility risk genes.

a, b Scoring for gene prioritisation procedure. Each column (b) represents a scoring criterion, with a shaded row denoting that said criterion is satisfied for the genes listed by each row. The criteria were as follows: closest gene to lead SNP, gene mapped to 95% credible set for probabilistic finemapping of per variant effect sizes, gene prioritised by probabilistic finemapping of marginal TWAS Z (FOCUS), gene annotated with strongest eQTL in the locus, gene annotated with strongest pQTL in the locus, gene prioritised by the OpenTargets V2G pipeline, variant annotated to gene with highest CADD score, gene with non-synonymous variant in locus, and variant annotated to gene with lowest RegulomeDB rank score. An unshaded cell for any gene denotes that the gene did not satisfy that criterion. c Phenome-wide investigation of genetically proxied TNFRSF1A using a blood eQTL with high confidence for annotation as a causal variant. Mendelian randomisation leveraged this eQTL effect size as an IV to estimate the causal effect of TNFRSF1A expression on binary and continuous phenotypes in the IEUGWAS database. The results for traits that survive Bonferroni correction are visualised as Z scores (MR β/MR SE) and have been flipped relative to inhibition of the gene such that Z > 0 corresponds to decreased expression increasing the odds (binary) or value (continuous) of the outcome phenotype.