Systematic evaluation of methylation-based cell type deconvolution methods for plasma cell-free DNA

Abstract

Plasma cell-free DNA (cfDNA) is derived from cellular death in various tissues. Investigating the tissue origin of cfDNA through cell type deconvolution, we can detect changes in tissue homeostasis that occur during disease progression or in response to treatment. Consequently, cfDNA has emerged as a valuable noninvasive biomarker for disease detection and treatment monitoring. Although there are many methylation-based methods for cfDNA cell type deconvolution, a comprehensive and systematic evaluation of these methods has yet to be conducted. In this study, we benchmark five methods: MethAtlas, cfNOMe toolkit, CelFiE, CelFEER, and UXM. Utilizing deep whole-genome bisulfite sequencing data from 35 human cell types, we generate in silico cfDNA samples with ground truth cell type proportions to assess the deconvolution performance of the five methods under multiple scenarios. Our findings indicate that multiple factors, including reference marker selection, sequencing depth, and reference atlas completeness, jointly influence the deconvolution performance. Notably, an incomplete reference with missing markers or cell types leads to suboptimal results. We observe performance differences among methods under varying conditions, underscoring the importance of tailoring cfDNA deconvolution analyses. To increase the clinical relevance of our findings, we further evaluate each method’s performance in potential clinical applications using real-world datasets. Based on the benchmark results, we propose general guidelines to choose the suitable methods based on sequencing depth of the cfDNA data and completeness of the reference atlas to maximize the performance of methylation-based cfDNA cell type deconvolution.