Fig. 2. Performance of SAIGE-GENE+ in UKBB WES data.

a, Computation time and memory of the gene-based tests (Step 2; Methods) in SAIGE-GENE and SAIGE-GENE+ for four genes with different numbers of variants. The SKAT-O tests were conducted with three maximum MAF cutoffs (1%, 0.1% and 0.01%) and three variant annotations (LoF only, LoF+missense and LoF+missense+synonymous) and combined using the Cauchy combination or minimum P value approach. Plots are in the log₁₀ –log₁₀ scale. Details of the numbers and genes are presented in Supplementary Table 1. b, Most significant variant sets across the three different MAF cutoffs (1%, 0.1% and 0.01%) and three functional annotations (LoF (L) only, LoF+missense (M+L) and LoF+missense+synonymous (S+M+L)). Distribution of variant sets with the smallest P values among 551 significant gene–phenotype associations identified by SAIGE-GENE+ in the analyses of 30 quantitative traits and 141 binary traits in the UKBB WES data.