Manhattan plot for genome-wide burden analysis under the co-dominant (top) and recessive (bottom) models. For each gene, the negative log-transformed p value of the joint analysis for the most significant variant set under a co-dominant (top) or recessive (bottom) model is plotted. For each gene, variant sets providing inconsistent results across the joint analysis, the trans-ethnic meta-analysis, and the trans-pipeline meta-analysis (i.e., P < 0.001 in the joint analysis and P > 0.05 in the trans-ethnic or trans-pipeline meta-analysis) were discarded. The red lines represent the significance threshold after Bonferroni correction to account for the total number of independent tests ( P = 4.61 x 10 -7 under a co-dominant model and 1.85 x 10 -6 under a recessive model). The names of the top-ranked genes with a joint P < 10 -4 are shown in red for rare variants associated with an increase in the risk of critical COVID-19 and in blue otherwise
