GFOLD diff input

Question

[tonja.r]

I have normalized rna-seq count data by deseq and edger. I would like to use these already normalized data for GFOLD. However, there are three options for GFOLD considering the normalizations: -norm <Count/DESeq/NO>. In the paper they write:

n is a normalization constant reflecting the sequencing depth and l is the gene length. The method proposed by Anders and Huber (Anders & Huber 2010) was used to calculate n. If this method fails, we simply treat n as the library size (sequencing depth).

So, I am a bit concerned if I set -norm NO, the calculations would be different if I just use my raw data and use -norm DESeq

Answer 1

[jianxing.tongji]

Yes, different normalization gives different result. However, the top ranked genes should be similar.