I started to investigate TCGA data, in particular, the correlation between miRNAs and mRNAs. Considering TCGA Data Level 3 (mRNA, miRNA), both miRNA and mRNA data are provided already as normalized datasets, albeit in different ways.
My goal is to investigate correlation between miRNAs and mRNAs, reported as RPMs and RPKMs, respectively.
With this in mind, I would proceed by applying Spearman's Correlation (to identify negatively-correlated miRNA-mRNA pairs). Before proceeding though, I would like to understand if I need to apply any normalization before the correlation analysis (should I run the correlation on the RPMs and RPKMs, or should I normalize the raw data, i.e. by applying TMM, and only then do the correlation analysis)?
Thanks for any suggestion.