We routinely work with gene-expression data sets generated from whole-blood of subjects of case and control status. Recently, there seems to be quite an interest in deconvolving cell-type specific gene-expression signatures from whole blood and we too think this is an interesting concept.
Numerous programs available: CellMix csSAM DeMiX iSOLATE PERT DeconSeq (for RNA-Seq) and I am sure there are others...
I have three main questions:
For now we are interested in exploring the utility of CellMix and csSAM. I am under the impression that CellMix can deconvolve cell-type signatures from whole-blood with out any prior knowledge of cell-type frequencies where csSAM deconvolves known cell-types signatures (and measures strength of differential expression between cell-tyes). Is this a correct assessment?
Described usage of these programs is slightly ambiguous. Can someone point me towards towards working example (as in a proper usage guid) of deconvolving cell-type contributions to whole blood gene-expression of samples with unknown cell-types using CellMix and csSAM? I find the below links very vague descriptions of how to perform this on local data-sets. In other words, starting with a gene-expression matrix derived from whole-blood and estimating cell-type proportions and using csSAM to perform differential expression on these cell-types across two conditions. Any help is much appreciated.
For this specific case, do we think that an analysis using first CellMix to deconvolve whole-blood cell-types following by csSAM to be an effective pipeline?