Fig. 4From: Metabolomic predictors of phenotypic traits can replace and complement measured clinical variables in population-scale expression profiling studiesMeta-analyses and replication studies. Number of meta-analyzed genes (significant associations, bacon-adjusted p-values FDR-adjusted for multiple testing, p<0.05) in leave-one-cohort meta-analyses (A) and percentage of genes replicated (significant associations, Bonferroni-adjusted for multiple testing, p<0.05) in replication cohort (B). Type of outcome variable is indicated by color: reported or measured variable = black, metabolomic surrogate = orange. Mean values across four meta-analyses/replication studies are plotted as horizontal bars. Note the log10 scale on the y-axis of the upper plotBack to article page