Dear all, I have expression counts from an RNAseq experiment for several libraries. I now want to use the MDS plot function to visualise similarities of the expression profiles of the various libraries. Looking at the expression matrix as well as at the resulting MDS plot, however, I saw some aspects different to my expectation, so I tested some aspects on a small example, e.g.: c<-cbind(a=rep(2,50),b=rep(4, 50),c=rep(8,50),d=rep(8,50) ) d <- DGEList(counts=c) jpeg(filename="mds-plot_matrix-abcd",width=700,height=700) plotMDS(d) dev.off() c<-cbind(b=rep(4,50),c=rep(8,50),d=rep(8,50),a=rep(2,50) ) d <- DGEList(counts=c) jpeg(filename="mds-plot_matrix-bcda",width=700,height=700) plotMDS(d) dev.off() With those results, however, I got totally confused: - why don't identical libraries cluster together? - why does the positioning of the libraries in the mds plot depend on the ordering of the columns in the expression matrix? What did I do wrong here? Thanks a lot, Lina [[alternative HTML version deleted]]