Fig. 4

Validation of the unreliability method on datasets with technical replicates. Association of type II probe variability estimated by the averaged, absolute methylation differences between repeated samples (∆β) with a the unreliability score and b the mean intensity (MI) in the TACT study (n = 2 × 71 longitudinal paired blood samples), the Repeatability set (n = 4 × 4 × 4 technical replicates for fresh blood, frozen blood and cervical smear samples) and GSE174422 (n = 2 × 128 technical replicates for blood samples). For type I probes see Supplementary Figs. S10–S11. c Distribution of Δβ for all type II probes (grey) or those removed by the p-value detection method at different threshold settings (red, orange, yellow) and the unreliability method (black) in the different datasets. d Correlation of MI measured for type II probes across datasets. e Mean red versus green total signal intensities of the samples in the different datasets. f Noise distributions obtained from the negative control probes in the different datasets