A Robust Bayesian Approach to Bulk Gene Expression Deconvolution with Noisy Reference Signatures

Abstract

Differential gene expression in bulk transcriptomics data can reflect regulated change of transcript abundance within a cell type and/or change in the proportion of cell types within the sample. To differentiate these scenarios, bulk expression deconvolution methods have been developed, which reveal cell type proportions and transcriptomes at the larger scales afforded by bulk RNA-seq compared to single-cell RNA-seq. However, the accuracy of these methods is highly sensitive to technical and biological differences between bulk profiles and the cell type-signatures required as references during deconvolution. We present BEDwARS, a Bayesian deconvolution method specifically designed to address potential differences between reference signatures and true but unknown signatures underlying the bulk transcriptomic profiles. Through extensive benchmarking utilizing eight different datasets derived from pancreas and brain, and by generating additional noisy reference signatures, we demonstrate that BEDwARS outperforms leading in-class methods for estimating cell type proportions and is more robust to noise in reference signatures. Furthermore, it more accurately estimates true cell type signatures compared to the state-of-the-art method. Application of BEDwARS to newly generated RNA-seq and scRNA-seq data on a rare pediatric condition (Dihydropyridine Dehydrogenase deficiency) revealed the possible involvement of ciliopathy and impaired translational control in the etiology of the disorder.

imply: improving cell-type deconvolution accuracy using personalized reference profiles

“…This assumption ignores person-to-person heterogeneity for cell-type-specific gene expression, and deviates from the biological fact that the gene expression profile could vary, even for one purified cell type, depending on environmental influences, age, sex, subject’s health status, and treatment paradigms (Kedlian et al, 2019; Di Biase et al, 2022; Findley et al, 2021; Idaghdour et al, 2010; Aguirre-Gamboa et al, 2016; Gibson, 2008; Çalişkan et al, 2015; Troester et al, 2004; Modlich et al, 2004). Mismatched reference signatures can impact the deconvolution results (Sutton et al, 2022; Ghaffari et al, 2023). The problem is even exacerbated when handling longitudinally observed and repeatedly-measured data, when intra-subject samples share information and inter-subject heterogeneities are relatively strong.…”

“…Mismatched reference signatures can impact the deconvolution results (Sutton et al, 2022;Ghaffari et al, 2023). The problem is even exacerbated when handling longitudinally observed and repeatedly-measured data, when intra-subject samples share information and intersubject heterogeneities are relatively strong.…”