Fig. 6: Decomposition of bulk DNA methylation data into latent methylation components.
From: Evidence of DNA methylation heterogeneity and epipolymorphism in kidney cancer tissue samples
Schematic explaining how bulk DNA methylation data are decomposed into cell-type-specific latent methylation components (LMCs) and the proportion of LMCs in each sample (A). Heatmap demonstrating the 7 LMCs, and the proportion of each of these LMCs in the tissue samples (B). For each sample, the top annotation bar shows the pathology (ccRCC vs normal kidney), sample purity and patient ID (from which the sample was derived). Heatmap demonstrating correlation values (cor) for the proportion of each LMC and purity values, for tissue samples (C). Purity values are derived using three independent methods: WES, RNA-seq (‘ESIMATE’) and DNA methylation (‘InfiniumPurify’). ‘ESTIMATE’ is the only method which provides purity estimates for normal samples, therefore these are shown separately in the heatmap. Heatmap demonstrating correlation values (cor) between each LMC and reference methylomes for various cell types (D). For each cell type, the accompanying code denotes the reference from which it was derived (from publicly available reference methylomes). In (C, D) a significant positive and negative correlation are shown in red and blue respectively. White denotes no significant correlation (p value > 0.05). Cluster dendrogram obtained using methylomes for the seven LMCs and reference methylomes (E). LMC1 and LMC3 cluster with immune cells, on the left branch of the dendrogram.
