Skip to main content
Fig. 3 | BMC Bioinformatics

Fig. 3

From: A comparison of reference-based algorithms for correcting cell-type heterogeneity in Epigenome-Wide Association Studies

Fig. 3

DHS information improves quality of the reference database. a Using 100 in-silico mixtures of 5 purified blood cell subtypes from Zilbauer et al., we compare the R2 values (R2, y-axis) of the estimated cell proportions as obtained using EpiDISH between the two different reference databases: one which uses cell-type specific DHSs to select DMCs when constructing the DNAm reference centroids (DHS), and another which only uses DMCs regardless of DHS status (noDHS). The average R2 value over all cell-types is being shown. A total of 25 different Monte Carlo runs were performed to obtain 25 average R2 values (x-axis) for each reference database. P-value is from a one-tailed paired Wilcoxon-rank sum test. (b) As (a) but now 100 in-silico generated mixtures of 3 epithelial cell subtypes (breast, renal cortical and pancreas) for which DHS data was available. (c) As (b), but now for 3 other cell subtypes (fetal lung fibroblast-IMR90, hepatocytes and B-cells) with available DHS information and DNAm profiles from two independent studies

Back to article page