A Critical Survey of Deconvolution Methods for Separating Cell Types in Complex Tissues

Abstract

Abstract-Identifying properties and concentrations of components from an observed mixture, known as deconvolution, is a fundamental problem in signal processing. It has diverse applications in fields ranging from hyperspectral imaging to denoising readings from biomedical sensors. This paper focuses on in-silico deconvolution of signals associated with complex tissues into their constitutive cell-type specific components, along with a quantitative characterization of the celltypes. Deconvolving mixed tissues/cell-types is useful in the removal of contaminants (e.g., surrounding cells) from tumor biopsies, as well as in monitoring changes in the cell population in response to treatment or infection. In these contexts, the observed signal from the mixture of cell-types is assumed to be a convolution, using a linear instantaneous (LI) mixing process, of the expression levels of genes in constitutive celltypes. The goal is to use known signals corresponding to individual celltypes along with a model of the mixing process to cast the deconvolution problem as a suitable optimization problem.In this paper, we present a survey and in-depth analysis of models, methods, and assumptions underlying deconvolution techniques. We investigate the choice of the different loss functions for evaluating estimation error, constraints on solutions, preprocessing and data filtering, feature selection, and regularization to enhance the quality of solutions, along with the impact of these choices on the performance of commonly used regression-based methods for deconvolution. We assess different combinations of these factors and use detailed statistical measures to evaluate their effectiveness. Some of these combinations have been proposed in the literature, whereas others represent novel algorithmic choices for deconvolution. We identify shortcomings of current methods and avenues for further investigation. For many of the identified shortcomings, such as normalization issues and data filtering, we provide new solutions. We summarize our findings in a prescriptive step-bystep process, which can be applied to a wide range of deconvolution problems.

High-throughput genomic profiling of tumor-infiltrating leukocytes

“…However, if a small number of lineage specific genes are already known, then it might be possible to derive additional marker genes using in silico nanodissection, a novel machine learning technique for predicting cell type-specific genes from GEP mixture data [44]. We refer the reader to [16] and [39] for additional details of gene expression deconvolution methods, including marker gene selection methods and technical considerations. Table 1 summarizes various GEP enrichment and deconvolution methods, highlighting and comparing their key features.…”

“…A biological mixture m can then be modeled as a system of linear equations in which m = G × f , where f is a vector containing the fraction of each cell subset from G in m . While methods have been proposed to determine f , G or both [16,17,21,22,24–30,32,35–37,39,41,49], many approaches estimate f given m and G . Here, we focus on deconvolution techniques that can enumerate TIL proportions ( f ).…”

“…For example, to profile a large diversity of cell types, Altboum and colleagues introduced a deconvolution approach based on elastic net regularization called Digital Cell Quantifier (DCQ) [36], and observed improvement through regularization. Indeed, regularization was found to improve the performance of several distinct deconvolution methods [39], suggesting the general utility of this approach.…”

Comprehensive Evaluation of RNA-seq Quantification Methods for Linearity

“…Deconvolution is a mathematical process used to extract constituent elements from a mixture of multiple signals [10]. In the field of biomedical research, deconvolution is widely applied to retrieve cell-type or tissue specific gene expression profiles from heterogeneous tissue samples.…”

“…In the field of biomedical research, deconvolution is widely applied to retrieve cell-type or tissue specific gene expression profiles from heterogeneous tissue samples. Most deconvolution algorithms in the literature assume a linear model [10][11][12][13][14][15][16][17], in which the expression signal of the mixture is a weighted sum of the expression for its constitutive cell types. Previous analysis has shown the necessity of using anti-log expression microarray data to avoid unwanted bias introduced by non-linear transformation [18].…”