a Table of cancer types for which publicly available DNA-methylation data were used. The coloring highlights pairs of neuroendocrine or epithelial cancers. b PCA on DNA-methylation beta values of SCLC, LUAD, NB, and GBM cell lines from (a). Shown is the amount of variation explained by each component independently in percent. The elbow at about 10 components indicates that later components are less informative. c PCA plot corresponding to (b) using PC2 and PC4. Each dot belongs to a single cell line and coloring refers to the cancer entities (left panel), neuroendocrine cancers (middle panel), or epithelial cancers (right panel). d CpGs ranked by their loadings of PC2 (left) and PC4 (right). Most informative CpGs are located at the curve tails, which show a sharp increase in loading values (indicated by black horizontal lines). The color code reflects the mean beta value for DNA-methylation of CpGs averaged over all neuroendocrine cancer cell lines (NB and SCLC) for PC2 and all epithelial cancer cell lines (LUAD and SCLC) for PC4. A strong blue coloring means that CpGs are low on methylation (vice versa for red coloring). e Gene Ontology analysis based on genes that correspond to CpGs selected in (d). Enrichment was performed independently for genes associated with PC2 (left) and genes associated with PC4 (right). f Neuroendocrine score based on PC2 (left panel) and the epithelial score (right panel) based on PC4. The scoring model was calculated by using CpG loadings selected in (d) and projected onto additional cell lines. Scores were Z-score normalized and higher values are associated with increased neuroendocrine or epithelial properties. Boxplots summarize the score distributions per cancer type (EWS Ewing’s sarcoma, FSARC fibrosarcoma, GBM glioblastoma, LCLC large cell lung cancer, LUAD lung adenocarcinoma, NEGC neuroendocrine gastric carcinoma, NB neuroblastoma, SCC squamous cell carcinoma, SCLC small cell lung cancer, UV-MCC UV-light-associated MCC, VP-MCC virus-associated MCC); each dot represents a single cell line.