Fig. 2From: Feature selection and causal analysis for microbiome studies in the presence of confounding using standardizationSimulation results Mean estimated false discovery proportion (FDP) for \(n = 100\) and Poisson features at varying nominal false discovery rate (FDR) values using Benjamini–Hotchberg adjusted p-values based on the debiased LASSO estimate following iterative sure independence screening (iterative SIS). The \(y = x\) line is shown in black; any values above this line indicate lack of FDR controlBack to article page