Deconvolving RNAseq data

[anglabat]

Hi,

I'm trying to deconvolve RNAseq data derived from mixed blood leukocytes by known cell proportions. Some of this data comes from model organisms, but other data comes from rare species. As such, I'm having difficulty applying any method that requires a matrix of expression levels of known cell type biomarkers. The way I see it, I have two choices

1) deconvolve everyone based on data for one model organism and discuss the data in that specific light. In this case I need to be pointed to a dataset/database of blood cell-specific biomarkers for human or mouse.

2) deconvolve using some other method that can use my known proportions, does not require me to add a matrix of biomarker expression.

I'm very new to this technology. Please forgive my ignorance. I'm happy for any advice.

[dpryan]

Have you considered just doing ICA? Particularly if you know approximately how many cell-types there are then that should be relatively straight forward.