Fig. 1: Selection and characterization of fCpG loci.
From: Fluctuating DNA methylation tracks cancer evolution at clinical scale
a, Schematic representing the study design. Bulk cancer tissue samples were collected, DNA extracted and methylation arrays were performed. Evolutionary dynamics were inferred and these were correlated with clinical variables and outcomes. The illustrations of the body, tumour, eppendorf tube and machine were reproduced courtesy of NIAID. The illustration of the physician was adapted from Science Figures, under an Open Design Licence 1.0. The illustration of the screen was created by Simon Dürr under a CC0 1.0 licence. b, fCpG methylation patterns reflect population evolutionary dynamics. Three scenarios are depicted: in a polyclonal population with a distant MRCA, diverse fCpG methylation (unmethylated in white, heterozygous in grey and methylated in black) results in a unimodal distribution of average methylation near the steady state (top). Following a recent clonal expansion (cell number 3 of the top panel), identical fCpG methylation across cells yields a characteristic ‘W-shape’ in the bulk methylation distribution (middle). Post-bottleneck, ongoing fluctuations generate diverse fCpG methylation patterns, changing the distribution of bulk methylation values (bottom). c, A hierarchically clustered heatmap of the 978 fCpGs identified in lymphoid cells (n = 2,204 samples). Magnified regions represent the homogeneous intermediate methylation pattern in normal lymphoid cells and the speckled pattern in cancer samples. PBMC, peripheral blood mononuclear cell; MGUS, monoclonal gammopathy of undetermined significance; MM, multiple myeloma; NOS, not otherwise specified; RT, Richter transformation. d, Example histograms of fCpG methylation distributions from healthy and neoplastic lymphoid cells (CLLs). e,f, Heatmaps showing the log2 fold change (FC) in genomic locations (e) and chromatin status (f) of fCpGs compared with non-fCpGs. g, Comparison between fCpG-associated and non-fCpG-associated genes in a single CLL sample (left; P = 5.14 × 10−12, Wilcoxon test, sample SCLL-328), and expression of fCpG-associated genes separated by discretized allele methylation status (right; P values were determined by two-sided Wilcoxon test, no multiple correction; n = 505 for fCpG genes and n = 15,736 for non-fCpG genes). The boxplot centre shows the median, the box shows the quartiles and whiskers represent ±1.5× interquartile range (IQR). TPM, transcripts per million.
