Figure 5

Effect of tumour purity on methylation signal in three TCGA cancer datasets. Plot (A) shows the distribution of the ABSOLUTE tumour purity estimates for Uterine, Kidney and Lung cancer datasets. Plot (B) shows the correlation of ABSOLUTE tumour purity estimates with the methylation signal for the top ranked differentially variable CpG site in the Kidney dataset. Plot (C) shows the correlation of ABSOLUTE tumour purity estimates with the methylation signal for a CpG site that is classified as differentially variable, differentially methylated and has a large R-squared value in the Uterine cancer dataset. Plot (D) shows the number of CpGs that have R-squared values of at least 10% in the top 50,000 differentially variable CpGs. The black dashed line clearly indicates that less than 5% of the top 50,000 differentially variable CpG sites are explained by tumour purity.