Fig. 3

CCTN-based prediction of gene expression levels for cancer cell lines and tumor patients. Gene-specific correlations between predicted and originally measured gene expression levels of individual genes comparing CCTN including only significant edges (pink) to CCTN using all edges (blue). A greater proportion of positive correlations reflects a better predictive power. a Prediction quality for human cancer cell lines used to train CCTN. As expected, CCTN using all learned edges is better than CCTN with significant edges only. b–l Prediction quality of CCTN for tumor patients of 11 independent TCGA cohorts. CCTN including only significant edges reaches strongly improved predictions for the vast majority of cohorts in comparison to CCTN with all learned edges. See Additional file 1: Figure S5 for all cohorts. AML acute myeloid leukemia, BRCA breast invasive carcinoma, CCLE Cancer Cell Line Encyclopedia, CCTN cancer cell transcriptional regulatory network, GBM glioblastoma multiforme, HNSC head and neck squamous cell carcinoma, LUAD lung adenocarcinoma, LUSC lung squamous cell carcinoma, OV ovarian serous cystadenocarcinoma, sig. significant, SKCM skin cutaneous melanoma, TCGA The Cancer Genome Atlas, COAD Colon adenocarcinoma, STAD Stomach adenocarcinoma, THCA Thyroid carcinoma