Deconvolution RNA-seq based on Single-RNA-seq?

Question

[star]

I have bulk RNAseq expression data from heterogeneous samples (with different cell types). I would like to estimate the cell type proportions in each sample using the cell markers.

There is a new publication on bioRxiv (https://www.biorxiv.org/content/10.1101/2020.01.10.897116v1) comparing different bulk RNAseq deconvolution methods. I am wondering which of those approaches can be used if we don’t know the expression level of gene markers in each cell type? On the other hand, the gene markers are available but not their expression or LogFold change.

Thanks a lot in advance!