If I understand you correctly you may try: cnts.2cond=cbind(rowMeans(cnts.norm[ , samp.idx]),rowMeans(cnts.norm[ , ref.idx])) then use cnts.2cond instead of cnts.d when calling pathview function: pv.out.list <- sapply(path.ids2, function(pid) pathview( gene.data = cnts.2cond, pathway.id = pid, species = "hsa", out.suffix=out.suffix, kegg.dir= "gageOutput/gageHG19", kegg.native=T, both.dirs = list(gene = F, cpd = T))) Note this way, what are plotting the absolute expression not the expression changes, hence you need to set both.dirs = list(gene = F, cpd = T). HTH, Weijun -------------------------------------------- On Fri, 5/23/14, Aric wrote: Weijun, I am not sure if what I want to do is possible, but I think it is. I would like to plot in pathview splitting by experiment condition for each gene instead of splitting by sample. For example, I have twelve samples divided equally into control and non-control. I would like to have one native kegg plot for each significant pathway, with the genes split according to whether the gene is up or down overall in the control or non-control group. It seemed like I have to use pure gage analysis to do this, but I can only get it split by sample. cnts.d <- cnts.norm[ , samp.idx] - rowMeans(cnts.norm[ , ref.idx]) path.ids <- c("hsa04670 Leukocyte transendothelial migration", "hsa04612 Antigen processing and presentation") path.ids2 <- substr(path.ids, 1, 8) pv.out.list <- sapply(path.ids2, function(pid) pathview( ? ? ? ? ? ? ? ? ? ? ? gene.data = cnts.d, pathway.id = pid, ? ? ? ? ? ? ? ? ? ? ? species = "hsa", out.suffix=out.suffix, ? ? ? ? ? ? ? ? ? ? ? kegg.dir= "gageOutput/gageHG19", kegg.native=T)) If I replace 'gene.data = cnts.d' with 'cnts.d[, 1:2]' then I get splitting in the plot, but only between samples one and two. I would like it by experimental condition. Is this possible?