Fig. 8: Factors influencing sequencing-based deconvolution accuracy.
From: Benchmarking of methods for DNA methylome deconvolution
a Heatmap showing Spearman’s R2 values on 100 in silico mixtures for all deconvolution methods at variable numbers of marker regions. b Line and scatter plot showing Spearman’s R2 values (y-axis) on 100 in silico mixtures for a variable number of marker regions per cell type (x-axis, log-scaled) at different sequencing depths. c Line and scatter plot showing Spearman’s R2 values (y-axis) on 100 in silico mixtures for a variable number of marker regions (x-axis, log-scaled) per cell type (x-axis, log-scaled) for skewed 13.8 (Q1 = 3, Q3 = 23) and homogeneous 13.7 (Q1 = 7, Q3 = 19) 14x covered sequencing data. Data are represented as mean ± SEM. P-values were calculated using a linear mixed model with deconvolution methods as random effects. *P < 0.05. ‘X’ symbol represents missing values. Source data are provided as a Source Data file.
