BEDwARS: a robust Bayesian approach to bulk gene expression deconvolution with noisy reference signatures.

Area of Science:

• Computational biology
• Genomics
• Bioinformatics

Background:

• Bulk transcriptomics data can reflect changes in gene expression due to altered transcript abundance within cell types or shifts in cell type proportions.
• Differentiating these two scenarios is crucial for accurate biological interpretation.
• Existing expression deconvolution methods face challenges with noisy reference signatures.

Purpose of the Study:

• To introduce BEDwARS, a novel Bayesian deconvolution method.
• To enhance the accuracy of estimating cell type proportions and signatures from bulk transcriptomics data.
• To improve robustness against noisy reference signatures.

Main Methods:

• BEDwARS employs a Bayesian framework to model and deconvolve gene expression data.
• The method is specifically designed to handle discrepancies between reference cell type signatures and true underlying signatures.
• Performance was evaluated against leading deconvolution methods using simulated and real-world datasets.

Main Results:

• BEDwARS demonstrates superior robustness when dealing with noisy reference signatures compared to existing methods.
• The method achieves higher accuracy in estimating both cell type proportions and cell type-specific expression signatures.
• Application to dihydropyridine dehydrogenase deficiency revealed potential links to ciliopathy and impaired translational control.

Conclusions:

• BEDwARS is a powerful and robust tool for accurate cell type deconvolution in bulk transcriptomics.
• The method facilitates a more precise understanding of cellular contributions to observed gene expression patterns.
• BEDwARS provides novel insights into the molecular underpinnings of dihydropyridine dehydrogenase deficiency.