Hey guys,

I've been flicking through some old data recently and i wanted a consensus on my approach.

I had a TPM normalised RNA-Seq gene count matrix, which I then filtered based on a pre-determined threshold. Now I'm left with those genes which meet my criteria which is great.

The question is; does one carry out another normalization on the remaining genes (given that potential artefacts/ owly expressed genes were removed), or do they go with the original normalisation, filtering of genes then downstream analyses?

I usually use EdgeR and deseq for these things, but this dataset I'm working on is old and was n=1. More of an exploratory works to find some useful candidates.

Thanks.