Journal: Nucleic Acids Research

Article Title: Dr.Nod: computational framework for discovery of regulatory non-coding drivers in tissue-matched distal regulatory elements

doi: 10.1093/nar/gkac1251

Figure Lengend Snippet: In solid cancers, oncogenes are enriched in driver-upregulated genes and TSGs are enriched in driver-downregulated genes. ( A ) The 52 target genes of the regulatory driver candidates, sorted by the size effect of their differential expression between the mutated and not mutated samples: 48 genes are driver-upregulated (positive y-axis), and 4 genes are driver-downregulated (negative y-axis). The colour of the bars represents the evidence in the literature for tissue-matched role of the genes as oncogenes (red colour) or tumour-suppressor genes (TSGs) (teal colour), with dark colour representing strong evidence (level 3 or 4) and light colour representing weak evidence (level 1 or 2; see Methods). Three genes had tissue-matched evidence in both directions (depending on phosphorylation, isoform, cellular localisation, disease stage etc.), which is denoted as two red-teal triangles. The circles below/above the bars represent the pan-cancer Cancer Gene Census (CGC) annotations: oncogenes (red), TSG (teal) or fusion (purple). The hearts and crosses represent the tissue-matched survival prognostic results based on The Human Protein Atlas : high expression predictive of favourable prognosis (teal hearts) or unfavourable prognosis (red crosses). The level of significance is denoted by the brightness of the colour. The y-axis (expression size effect) corresponds to the beta value of the MUT predictor in the Poisson GLM. In other words, it represents the log(FPKM-UQ) expression increase (or decrease if negative) in mutated samples, after accounting for copy-number variation. ( B ) Genomic view of an example gene EZH2 . Four regulatory regions (ABC enhancers) are shown, each in a different colour: purple in the EZH2 promoter, dark orange nearby the promoter, blue in intron of EZH2 , and a yellow enhancer 59 kb upstream of EZH2 . The yellow enhancer has seven mutations, shown as red stars in the inset below (two are in one sample). ( C ) The EZH2 expression in samples with (red) and without (grey) mutations in the regulatory regions of EZH2 . ( D ) The Kaplan–Meier plot of survival probability in samples with high (pink) vs. low (blue) EZH2 expression, with the log-rank test P -value shown on top, data from The Human Protein Atlas . ( E ) Examples of expression in five driver-upregulated genes and one driver-downregulated gene. ( F ) The average dependency score across 939 cancer cell lines from the DepMap Achilles project in the driver-upregulated genes in solid cancers (red) and the control genes (all genes not regulated by the regulatory driver candidates), with two-tailed Wilcoxon rank-sum test P -value shown on top. The dependency score represents how dependent the cell line is on the gene, i.e. how essential the gene is for viability of the cell line based on CRISRP/Cas9 screen.

Article Snippet: To compare gene expression between the donors with vs without high-CADD mutations in the regulatory space of the gene, we modelled the gene expression using Poisson Generalised Linear Model (GLM) regression as FPKM-UQ ∼ MUT + CNV, using MATLAB function fitglm(data, ‘linear’, ‘Distribution’, ‘poisson’, ‘DispersionFlag’, true).

Techniques: Quantitative Proteomics, Phospho-proteomics, Expressing, Control, Two Tailed Test