Skip to main content
Fig. 2 | BMC Bioinformatics

Fig. 2

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

Fig. 2

Validation of EpiDISH in independent data with FACS and known mixture estimates. a Left panels: Barplots of the average weight proportions of each major blood cell subtype according to FACS and EpiDISH, in 354 rheumatoid arthritis (RA) cases and 335 controls. Right panel depicts the difference between cases and controls. b Left panel: Barplots of the average weight proportions of each major blood cell subtype according to FACS and EpiDISH in 6 whole blood samples from Koestler et al. Root Mean Square Error (RMSE) is given. Right panel: Scatterplot of the corresponding estimated cell fractions vs the FACS estimates for all 6 samples and each cell subtype. R2 values (rounded to two significant digits) for each cell subtype are given, as well as the average value over all cell subtypes. c Scatterplots comparing the estimated cell fractions using EpiDISH to the true known fractions for 12 experimentally reconstructed “whole blood” samples, where the mixing proportions were known. One scatterplot is shown for each blood cell subtype used in the reconstructions. In each case, we give the RMSE and R2 value (latter values have been rounded to two significant digits)

Back to article page