Good morning all,

I was wondering if it's possible to use CIBERSORT (or similar deconvolution algorhitm) on RNAseq data of gut biopsies from humans?

We are trying to gauge the presence of immune cell types in them, but the biopsies also contained epithelia, fibroblasts and other cells. I am worried that using the LM22 signature set will give me bias due to many transcripts coming from these non-hematopoietic cell types. Based on what I understand, this should be possible, right?

Second part of my question, we get quite high (insignificant) deconvolution p-values. Is there any way to improve them?

My example outcome attached.

Thank you very much for your time,

Adam