The extent of stromal and immune cell infiltration within solid tumors has prognostic and predictive significance. Unfortunately, expression profiling of tumors has, until very recently, largely been undertaken using bulk techniques (e.g., microarray and RNA-seq). Unlike single-cell methods (e.g., single-cell RNA-seq, FACS, mass cytometry, or immunohistochemistry), bulk approaches average expression across all cells (cancer, stromal, and immune) within the sample and, hence, do not directly quantitate tumor infiltration. This information can be recovered by computational tumor deconvolution methods, which would thus allow interrogation of immune subpopulations across the large collection of public bulk expression datasets. The goal of this Challenge is to evaluate the ability of computational methods to deconvolve bulk expression data, reflecting a mixture of cell types, into individual immune components. Methods will be assessed based on in vitro and in silico admixtures specifically generated for this Challenge.