Question from a user:

I have a few questions in relation to differential TE expression, especially when using per TE outputs from tools like Telescope [https://github.com/mlbendall/telescope] (in my case, a Telescope based tool [https://github.com/jasonwong-lab/LocusMasterTE]). As the tool suggests, I use the final_count output from Telescope. For normalization, my understanding is that DEseq2 bases its calculation on the total counts in the input file. I however expect there to be global changes in TE expression, hence cannot solely rely on normalization based on TE counts only.

I wanted to ask if it is possible to perform library size normalization in DEseq2 such as the following, where the library size vector corresponds to the total number of aligned reads. Or would this interfere with DESeq’s downstream model?

DESeq2 normalization uses the median ratio of a sample's count to a reference as the size factor. You can also specify the features over which to compute this median use controlGenes. But it relies on having a set of genes where this provides a reasonable vector for scaling. If all counts go up, because of global up-regulation, this is not inferrable from an assay that captures relative abundance.