Fig. 1

Identification of metabolic transcriptional components (mTCs). A Workflow for identification of mTCs. Consensus independent component analysis (c-ICA) is applied to identify transcriptional components (TCs). Subsequent systematic selection of TCs enriched for metabolic processes resulted in 132, 151, 136, and 136 mTCs for the GEO, TCGA, CCLE, and GDSC datasets, respectively. B Hierarchically clustered heatmaps showing the enrichment of the 608 metabolic gene sets of mTCs identified in the GEO, TCGA, CCLE, and GDSC datasets. C Scatter plot showing absolute Spearman correlation coefficients (x-axis) versus the percentage of overlapping top genes (genes with absolute weight > 3) between GEO mTCs and TCGA mTCs (y-axis). Only significant pair-wise correlations (with P-values < 0.05) are shown. Colored dots show the correlations > 0.5, the size of the dots represents the P-value of these Spearman correlations. The transparency of the dots is the same for all data points. Darker dot colors therefore mean that multiple data points are overlapping. D Venn diagram quantifying the overlap of mTCs between each dataset based on their pair-wise correlations. Two mTCs are counted as shared between datasets, when they have a high absolute Spearman correlation (|r_s| > 0.5). Three groups of (shared) mTCs, mentioned in the text, are designated