Hi! So, I am doing a co-expression analysis on DEGs for a TCGA dataset.
I have used raw count data to find DEGs with DESeq2 and have around 300 of them. Now, I would like to proceed with co-expression analysis: should I normalize the raw count data I have or can I extract the DEGs from my original dataset (again, raw counts) and compute the correlation matrix between those? Or should I go back and download the FPKM data (which I know is normalized) and extract from that the DEGs I computed by providing the raw counts to DEseq2?