Heatmap of contributions of a select subset of biomarkers to the variable importance measure changes as assessed by influence curve-based estimation, across all subjects. The heatmap produced performs supervised clustering, as per Pollard & van der Laan (2008) <doi:10.2202/1544-6115.1404>.

heatmap_ic(x, ..., design, FDRcutoff = 0.25, type = c("top", "all"), top = 25)



Object of class biotmle as produced by an appropriate call to biomarkertmle.


additional arguments passed to superheat::superheat as necessary


A vector giving the contrast to be displayed in the heatmap.


Cutoff to be used in controlling the False Discovery Rate.


A character describing whether to plot only a top number (as defined by FDR-corrected p-value) of biomarkers or all biomarkers.


Number of identified biomarkers to plot in the heatmap.


heatmap (from superheat) using hierarchical clustering to plot the changes in the variable importance measure for all subjects across a specified top number of biomarkers.


if (FALSE) { library(dplyr) library(biotmleData) library(SummarizedExperiment) data(illuminaData) colData(illuminaData) <- colData(illuminaData) %>% data.frame() %>% mutate(age = as.numeric(age > median(age))) %>% DataFrame() benz_idx <- which(names(colData(illuminaData)) %in% "benzene") biomarkerTMLEout <- biomarkertmle( se = illuminaData, varInt = benz_idx, bppar_type = BiocParallel::SerialParam(), g_lib = c("SL.mean", "SL.glm"), Q_lib = c("SL.mean", "SL.glm") ) limmaTMLEout <- modtest_ic(biotmle = biomarkerTMLEout) heatmap_ic(x = limmaTMLEout, design = design, FDRcutoff = 0.05, top = 10) }