Hi, I have asked a question here few months back Multi factor design edgeR/Deseq2

I am just trying to learn more rna-seq statistics and R, so I am playing around with the data in edgeR. I have paired samples. When I do plotMDS, few of the paired samples appears very close to each other i.e ( indvidual_1_treated, indvidual_1_Untrt ) ( indvidual_2_treated, indvidual_2_Untrt ) etc. Does it mean the effect of drug is not shown any difference in that individual ? How to interpret this ?

library(edgeR)

raw_counts <- read.table("raw_count_matrix.txt", header=TRUE,sep="\t", row.names=1, as.is=TRUE)
filter <- apply(raw_counts, 1, function(x) length(x[x>5]) >=6)
filtered <- raw_counts[filter,]

treat <- as.factor(rep(c("Treat","Untrt"), each=12))
patient <- as.factor(c(1:12, 1:12))
design <- model.matrix(~patient+treat)

y <- DGEList(counts=filtered)

y <- calcNormFactors(y)
y <- estimateGLMCommonDisp(y,design)
y <- estimateGLMTrendedDisp(y,design)
y <- estimateGLMTagwiseDisp(y,design)

plotMDS(y)

So you're saying that the treated and untreated samples for a given individual cluster closely together? If so, that doesn't necessarily mean that there's no drug effect for that indivudal. The MDS plot is only showing you the largest 2 dimensions of the data, and it's possible that the drug effect is visible in the 3rd, 4th, or another smaller dimension, You can pass the dim.plot argument to plotMDS to plot other dimensions. And even if you don't find any dimension in the MDS plot that clearly separates treated from untreated, there may still be a few differentially expressed genes anyway.