Are the deconvolved gene expression values (already in fraction) from Cibersotx normalized or require any further normalization like TMM?

Question

[Researcher]

Hi all, i am using Cibersotx to deconvolve TCGA gene expression data and getting my the expression values in fractions. I am not sure whether these should be considered as normalized count to perform differential expression among samples or we require any normalization as TMM in EdgeR. Anybody has any experience with this? Thanks in advance.