I just finished spending weeks setting up and running alignment and quantification on RSEM for my RNA reads. I am getting ready to perform DEA, starting with normalization. However, I learned that RSEM - using STAR as the aligner- assumes Transposable Elements are silent (TEs), hence it does not account for them. I intend to account for TEs, so I found SQUIRE, which does account for TEs. I am not sure if I have to start my entire alignment and quantification process from scratch to ensure my pipeline is clean, or if I should just run SQUIRE for TEs and then merge the results with my gene quantification data from RSEM.

I am concerned about compatibility and library size biases when performing normalization and other downstream differential expression analysis tasks. Any help, please?