Comparing Gene Expression Across Cell Types

[Juanmu-Bio]

Hi, and thank you for developing BayesPrism, it is an extremely useful tool.

I followed the deconvolution tutorial and obtained the Z matrix (bulk × gene × cell type).
Now I would like to associate specific genes with specific cell types, and for that I need to compare the expression of the same gene across different cell types using the deconvolved expression matrix Z.

However, I am unsure what is the recommended and statistically valid way to perform such comparisons.

So my question is:
What is the appropriate method to compare the expression of a given gene across cell types using BayesPrism outputs?
If differential expression analysis is not recommended, is there an alternative approach suggested by the authors?

Thank you in advance, and apologies if this is a basic question — I am still new to bioinformatics :)

[tinyi]

Thank you for your question. You may simply round up Z and use it as input for DESeq2.

…

On Wed, Dec 3, 2025 at 6:37 AM Juanmu-Bio ***@***.***> wrote: *Juanmu-Bio* created an issue (Danko-Lab/BayesPrism#124) <#124> Hi, and thank you for developing BayesPrism, it is an extremely useful tool. I followed the deconvolution tutorial and obtained the Z matrix (bulk × gene × cell type). Now I would like to associate specific genes with specific cell types, and for that I need to compare the expression of the same gene across different cell types using the deconvolved expression matrix Z. However, I am unsure what is the recommended and statistically valid way to perform such comparisons. I thought about doing differential expression on the normalized pseudocounts (e.g., VST-transformed Z), but if I am not mistaken this may be misleading, because pseudocounts can lead to false positives. So my question is: What is the appropriate method to compare the expression of a given gene across cell types using BayesPrism outputs? If differential expression analysis is not recommended, is there an alternative approach suggested by the authors? Thank you in advance, and apologies if this is a basic question — I am still new to bioinformatics :) — Reply to this email directly, view it on GitHub <#124>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AB4NHSZEANLU45F5OIVQIAT373YTPAVCNFSM6AAAAACN5SBRQ6VHI2DSMVQWIX3LMV43ASLTON2WKOZTGY4TANJRHA4TENI> . You are receiving this because you are subscribed to this thread.Message ID: ***@***.***>