Hi Chris, Sorry for delay, I was out all of last week. The plots you are seeing are comparing the empirical distribution of batch effects (black lines and points) with the parametric estimates (red lines) that ComBat uses if par.prior=T. The top plots are for the mean of the batch effect and the bottom plots are for the batch variances. In practice, you only need to use the non-parametric prior when you have a very large deviation from the parametric prior--extreme skewness in the black compared to the red, or some form of bimodality in the black. In most cases, the results are nearly identical within a 1-2% difference in the expression values. I would be surprised if you see much difference in "par.prior=T" compared to "par.prior=F" with your data, so go ahead and use the default approach. Here are a few prior bioconductor forum discussions on the topic: [combat] par.priot=FALSE ComBat - stuck please help! about ComBat plot for non parametric Let me know if this helps. Thanks! Evan On Aug 18, 2015, at 7:29 AM, chris85 [bioc] <noreply@bioconductor.org<mailto:noreply@bioconductor.org>> wrote: Activity on a post you are following on support.bioconductor.org<https: support.bioconductor.org=""/> User chris85<https: support.bioconductor.org="" u="" 8408=""/> wrote Question: Prior plots from COMBAT<https: support.bioconductor.org="" p="" 71198=""/>: Hi I have had my first go using COMBAT to correct for batch effects in our data. I did quantile normalisation first. I am wondering what the prior plots mean that are outputted, could anyone give me an explanation of what I should be looking for here, the black and red lines etc? Upper and lower plots, there are 4 in total. batch = des2$BATCH #vector of numbers refering to the different batches of the samples modcombat = model.matrix(~1, data=pheno) #yields a vector of 1s of length pheno matrix <- as.matrix(data) combat_edata = ComBat(dat=matrix, batch=batch, mod=modcombat, par.prior=TRUE, prior.plots=TRUE) combat_df <- data.frame(combat_edata) Thanks, Chris ________________________________ Post tags: limma, normalization, microarray You may reply via email or visit Prior plots from COMBAT