Description of the Method

This method originates from CellMix, a comprehensive toolbox for gene expression deconvolution1.

As a gene marker list input, it uses the CEllmatch2 database included in the scCATCH tool Science 2020).

Deconvolution

The algorithm consists in a semi-supervised NMF-based deconvolution using KL divergence.

We use the CellMix::ged function using the parameters: method = "ssKL", sscale = FALSE, maxIter=500, log = FALSE.

Results

The program returns an estimated proportion matrix and the corresponding estimated cell-type specific profiles matrix for selected gene markers.

Descriptors

Omics_type = trancriptome

Cancer_type = all

Method_type = semi-supervised

Cell_type = estimation of all cell types or immune cell types