I am comparing different normalisation methods for my count table (I'm doing RNAseq analysis). One of them is DESeq2, the other one is RUVs. One way to compare them is to look at the unwanted variation that is still in the dataset after normalisation. This can be visualised using RLE plots.
RLE plots can be made using the function plotRLE. The main argument for this function is the object to be plotted. This can be a numeric matrix or a SeqExpressionSet object containing the gene expression.
For the RUVs normalisation making an RLE plot is simple as the normalised object is a SeqExpressionSet. However, for the DESeqDataSet, it is more complicated. Could somebody tell me how to extract the count matrix with normalised counts from the DESeqDataSet?
For reference, my code to plot the RLE plot for the RUVs normalisation is the following:
library(RUVSeq) set <- newSeqExpressionSet(counts = as.matrix(counts), phenoData = sample_info) setUQ <- betweenLaneNormalization(set, which="upper") genes <- rownames(counts) replicates <- makeGroups(control) # control is a column of sample_info and therefore information from the SeqExpressionSet setRUVs9 <- RUVs(setUQ, genes, k = 9, replicates) plotRLE(setRUVs9)
Thanks a lot in advance. Felicitas