deconvolution estimate immune cell subsets

This study introduces a computational method for estimating the composition of immune cell subsets from bulk gene expression profiles of complex tissues, such as tumors. The method aims to improve existing deconvolution methods by using a new set of reference gene expression profiles that include more microarray replicates of immune cells frequently associated with poor cancer prognosis, such as T helper cells, regulatory T cells, and macrophage M1/M2 cells.

The core algorithm of the deconvolution method is ε-support vector regression (ε-SVR) with an L1-norm penalty loss function. The reference gene expression signature matrix for regression was constructed by selecting differentially expressed genes from 148 microarray-based gene expression profiles of 9 immune cell types using ANOVA and minimizing the condition number.