Schroeder, M. P., Rubio-Perez, C., Tamborero, D., Gonzalez-Perez, A. Intron retention is a widespread mechanism of tumor-suppressor inactivation. Analysis of TCGA datasets have mostly focused on somatic mutations and translocations, with less emphasis placed on gene amplifications. a log2 (chromatin peaks in promoters) for tumor suppressor and oncogenes detected in Pan-Cancer study, b boxplot showing log2 (chromatin peaks in introns), and c breast cancer log2 (chromatin peaks count). For example, Moonlight identified GAS7 as a hypermethylated tumor suppressor in lung cancer and as an hypomethylated oncogene in head-and-neck squamous cell tumors. Cancer aneuploidies are shaped primarily by effects on tumour - Nature Identifying these genes is a crucial step towards personalising treatment for cancer, but the complexity and diversity of cancerous cells make finding these genes difficult. Cancer genome landscapes. All authors read and approved the final paper. In a recent review, cancer progression was summarized across four different steps: cancer initiation, tumor propagation, metastasis to distant organs, and drug resistance to chemotherapy1. Moonlight identified three oncogenes, CMYA5, ASPM, and ERBB2, showing 34, 30, and 29 samples with missense mutations, respectively (Methods; Supplementary Data10). MapSplice: accurate mapping of RNA-seq reads for splice junction discovery. Lubanska, D. & Porter, L. Revisiting CDK inhibitors for treatment of glioblastoma multiforme. Increased MTHFD2 expression is associated with poor prognosis in breast cancer. This analysis detects multiple patterns of BPs when different conditions are selected. We used the Moonlight Process Z-score matrix as input to the random forest procedure, such that the BPs are the features that the learning method can include in the model. Labels around the plot specify the cancer type; the number of OCGs and TSGs for that cancer type are in parentheses. We supplied the output of the Moonlight Upstream Regulatory Analysis (Methods) to this model to score the biological processes. Abstract. Publishers note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. PubMed Central 19, 32103214 (2013). Nat. Cancer 4, 177183 (2004). Concomitant Notch activation and p53 deletion trigger epithelial-to-mesenchymal transition and metastasis in mouse gut. Biophys. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/. Identification of Druggable Cancer Driver Genes Amplified across - PLOS Science 339, 15461558 (2013). Interestingly, Moonlight detected GATA3 as an oncogene in breast cancer with several mutated samples: frameshift insertion, deletion, and splice site. Thus, Moonlight detected ANGPTL4 as a dual-role gene, a finding which was confirmed by a recent study38. 121, 27502767 (2011). A compendium of mutational cancer driver genes - Nature Colaprico, A., Olsen, C., Bailey, M.H. Concurrently, chromatin was more closed or had dampened signal for tumor suppressors. The upper panel shows boxplots of cell-line expression levels. Interestingly, Moonlight detected open chromatin peaks in the intron regions for tumor suppressors. Therefore, we identified differentially methylated regions between normal and tumor samples for 18 TCGA cancer types (Methods). We also showed the ability of Moonlight to identify associations between the aforementioned biological processes and the specific genes that regulate these processes. Harari, D. & Yarden, Y. Molecular mechanisms underlying ErbB2/HER2 action in breast cancer. Futreal, P. A. et al. Scalable open science approach for mutation calling of tumor exomes using multiple genomic pipelines. Tobelaim, W. S. et al. 17, 31553161 (2010). Moonlight identified the cell cycle kinase CDK4 as an oncogene in glioblastoma multiforme, with the highest normalized peak score (1164). 19, 53 (2017). Rev. This hypothesis-generating mechanism provides clues to which gene properties that can be confirmed using in vivo models such as patient-derived tumors xenografted in mice, or proliferation assays in cell culture. [version 2; peer review: 1 approved, 2 approved with reservations]. Chromatin accessibility impacts transcriptional reprogramming in oocytes. Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. "How do we stop cancer once it starts? observed amplification of oncogenes and deletion of tumor suppressors) (Supplementary Data9). The legacy level-3 data of the Pan-Cancer studies (18 cancer types), for which there were at least five samples of primary solid tumor (TP) or solid tissue normal (NT) available, were used in this study and downloaded in May 2018 from The Cancer Genome Atlas (TCGA) cohort deposited in the Genomic Data Commons (GDC) Data Portal (Supplementary Data4). 13, 14021428 (2013). A.C., C.O., C.C, T.T., T.C.S., A.V.O., and L.C. The genome is oriented horizontally from top to the bottom, and GISTIC q-values at each locus are plotted from the left to right on a log scale. In addition, 151 driver genes showed a dual-role effect (Fig. This data set represents the most uniform attempt to systematically provide mutation calls for TCGA tumors. However, we have limited ability to differentiate driver DNA methylation (DNAme) changes from passenger events. Furthermore, we extended our study to mutation data. Cancer Driver Genes Methods and Protocols Home Book Editors: Timothy K. Starr Includes cutting-edge methods and protocols Provides step-by-step detail essential for reproducible results Contains key notes and implementation advice from the experts Part of the book series: Methods in Molecular Biology (MIMB, volume 1907) 29k Accesses 30 Citations Comprehensive Characterization of Cancer Driver Genes and Mutations Oncogene 21, 54275440 (2002). Sci. A driver gene is one that contains driver gene mutations. Overall, Moonlight predicted 776 cancer driver genes (626 oncogenes and 150 tumor suppressors) in the analyses of breast-invasive carcinoma (Supplementary Data5). Nature 489, 7582 (2012). CAS USA 102, 1554515550 (2005). Cancer Inform. e5. Wei, P. et al. Cancer Res. CAS E.P.s group is supported by grants from LEO Foundation (grant number LF17006), the Innovation Fund Denmark (grant number 5189-00052B), and the Danish National Research Foundation (DNRF125). These findings were also supported by literature 22,73,74,75. assembled the display (figures and tables) items. 44, e71 (2016). The first one is the multi-class log-loss measure. Liu, F. et al. Abstract. Specifically, Moonlight identified BLC2 as an oncogene in thyroid carcinoma, through decreasing apoptosis and showing a peak in the exon region concurrently, confirmed by published data45. For example, Moonlight identified H2AF as a highly expressed oncogene in several breast-cancer cell lines. Hierarchical clustering was performed on the Euclidean distance matrix. Another example is apoptosis, which is generally downregulated in association with cancer progression. 2b) and hypothesize that genes with pattern (i) can act as oncogenes while genes with pattern (ii) can act as tumor suppressors. Antimicrob. Internet Explorer). Curr. Biophys. Genome Biol. Cancer Drivers Actionability Database (2014.12) This database contains data on the interactions with therapeutic agents an driver genes contained in Cancer Drivers Database (2014.12). designed and performed research and interpreted the data results. The list provided additional information such as the type of mutation, either dominant (448), recessive (134), dominant/recessive (7), or undeclared (3). Proc. RNA-seq raw counts of 7962 cases (7240 TP and 722 NT samples) aligned to the hg19 reference genome were downloaded from GDCs legacy archive, normalized, and filtered using the R/Bioconductor package TCGAbiolinks14 version 2.9.5 using GDCquery(), GDCdownload(), and GDCprepare() functions for tumor types (level 3, and platform IlluminaHiSeq_RNASeqV2), as well as using data.type as Gene expression quantification and file.type as results. Consequently, we speculate that a guided therapy of the mentioned drugs will be beneficial for breast-cancer treatment. Targeting the Clear Cell Sarcoma Oncogenic Driver Fusion Gene EWSR1::ATF1 by HDAC Inhibition. Hum. Google Scholar. We identify 299 driver genes with implications regarding their anatomical sites and cancer/cell types. The study showed that the epigenetic background of the cell type may only permit certain oncogenes or tumor suppressors to change roles. A dual role for the anti-apoptotic Bcl-2 protein in cancer: mitochondria versus endoplasmic reticulum. Significance Modern large-scale sequencing of human cancers seeks to comprehensively discover mutated genes that confer a selective advantage to cancer cells. PubMed Central With them, we classify cancer driver genes, unveiling their roles in cancer mechanisms. Introduction Cancer is a collection of diseases characterized by abnormal and uncontrolled cellular growth caused primarily by genetic mutations 1, 2. In this step of the analysis, we found 3390 genes that were differentially expressed (Methods, Supplementary Data3) when comparing normal and tumor breast-cancer tissue samples. Google Scholar. For the 3123 mediators predicted by Moonlight within 18 cancer types, 848 showed copy-number changes and 358 showed critical copy-number cancer driver genes (eg. TCGAbiolinks: an R/Bioconductor package for integrative analysis of TCGA data. J. Clin. Shen, L., Shi, Q. Epigenetics 7, 102 (2015). Supplementary Fig. Chen, X. et al. Haigis, K. M., Cichowski, K. & Elledge, S. J. Tissue-specificity in cancer: the rule, not the exception. Background Identifying the complete repertoire of genes that drive cancer in individual patients is crucial for precision oncology. 3b). Lobry, C., Oh, P. & Aifantis, I. Oncogenic and tumor suppressor functions of Notch in cancer: its NOTCH what you think. Cancer Cell 33, 690705 (2018). 183, 16451653 (2013). To this end, we present Moonlight, a tool that incorporates multiple -omics data to identify critical cancer driver genes. Certain cancer driver genes can exhibit oncogene or tumor-suppressor behavior depending on the biological context, which makes them difficult to identify. For breast cancer, we found that 231 (30%) of the predicted oncogenic mediators experienced epigenetic changes. Protoc. In Fig. We then sum over all classes (three in our case), adding the log value to the log loss if gene i belongs to class j according to the known truth. First, users can define the data type(s) incorporated for driver gene identification (the red rectangle in Supplementary Figure S1A ) and then select a specific dataset, for example, 'Glioblastoma multiforme' (GBM). Sci. We have characterized the specific molecular changes associated with all the 3123 oncogenic mediators and cancer driver genes in the following sections. Indeed, mutations can cause different effects such as a loss or reduction of mRNA transcripts impacting on the protein function. Nature Communications (Nat Commun) 11, 147156 (2012). In the parentheses, the number of OCGs and TSG for a specific molecular subtype; OCGs (green) and TSGs (yellow); purple and orange for mutations: inframe deletion, inframe insertion, missense; genegene edges between two cancer molecular subtypes are OCG in both (green), TSG in both (yellow), dual-role genes (red). Shao, B. et al. Takaku, M., Grimm, S. A. Furthermore, c-Kits dual-role behavior in different contexts has been already proposed63. Bioinformatics 30, 523530 (2014). 2e; Supplementary Data5). Several approaches have been developed to discover cancer driver genes and pathways, but these methods did not harness the power of integrating biological processes and their connection with gene deregulation to predict cancer driver genes12. The chromatin accessibility landscape of primary human cancers. Article Cancer Res. Tumour Biol. Mermel, C. H. et al. We assessed the results using two different quality measures, i.e., log loss and AUC (one class versus all). 9, 4421 (2018). Chen, H. et al. A census of human cancer genes. Moonlight predicted this gene to be an oncogene in renal clear-cell carcinoma associated with poor survival (log-rank test p=0.001, Fig. Front. Rep. 7, 985 (2017). Tamborero, D. et al. C.C. Cancer Genome Landscapes | Science While it has been shown that highly mutated genes promote cancer progression12, it is yet unknown if methylation and copy-number changes to cancer driver genes directly imply that these genes have been mutated. Cell 173, 305320 (2018). The main concept behind Moonlight relies on the observation that the classical approach to experimentally validated cancer driver genes consists in the modulation of their expression in cellular assays, together with the quantification of process markers, such as cellular proliferation, apoptosis, and invasion. Cell-of-origin patterns dominate the molecular classification of 10,000 tumors from 33 types of cancer. extended the original 20/20 rule2 in an ML approach allowing the integration of multiple ratiometric features of positive selection in 20/20+66 to predict oncogenes and TSGs from small somatic variants. Streicher, K. L., Yang, Z. Q., Draghici, S. & Ethier, S. P. Transforming function of the LSM1 oncogene in human breast cancers with the 8p11-12 amplicon. assessed the performance and accuracy of the method. Experimental validation confirmed 60%-85% of predicted mutations as likely drivers. Discov. PubMed Get what matters in cancer research, free to your inbox weekly. For example, we identified ADRA2A, predicted as oncogene in breast cancer and tumor suppressor in bladder urothelial carcinoma, targeted by 62 compounds. Branet, F., Caron, P., Camallires, M., Selves, J. 57, 1011 (2008). If the gene exhibits significant evidence after additional data integration, we define the genes that Moonlight discovered as cancer driver genes. Tseng, R.-C. et al. Clinical implication (a, b) KaplanMeier survival curves show that ANKRD23 is a tumor suppressor in BLCA (a) and an oncogene in KIRC (b). Annotation of cell lines were considered with TCGAs classification as reported in ftp://ftp.sanger.ac.uk/pub/project/cancerrxgene/releases/current_release/Cell_Lines_Details.xlsx. SOX17 regulates uterine epithelial-stromal cross-talk acting via a distal enhancer upstream of Ihh. e12. We then carry out a functional enrichment analysis computing a Moonlight Process Z-score that compares the literature-based knowledge to the result of the differential expression analysis. 44, e18 (2016). Med. 1b; Methods): (i) Moonlight identifies a set of Differentially Expressed Genes (DEGs) between two conditions, then (ii) the gene expression data are used to infer a Gene Regulatory Network (GRN) with the DEGs as vertices, and (iii) using Functional Enrichment Analysis (FEA), Moonlight considers a DEG in a biological system and quantifies the DEG-BP (biological process) association with a Moonlight Process Z-score. Mol. Porta-Pardo, E. et al. Our approach allows the interpretation of cancer-related pathways to identify essential cancer driver genes by integrating information on biological processes from literature with genegene interactions in transcriptomic data. 1 Altmetric Metrics Abstract Background Recent advances in next-generation sequencing technologies have helped investigators generate massive amounts of cancer genomic data. 4d, we report the results of the analysis from different mutation types for the cancer driver genes predicted by Moonlight in breast cancer. The Cancer Genome Atlas (TCGA) and Gene Expression Omnibus (GEO . J. Invest. In addition, it has been reported that mutation of different amino acid sites are related to antibiotic drug resistance77. 35, 86858690 (2014). Nucleic Acids Res. Genet. The total number of driver genes is unknown, but we assume that is considerably less than 19,000. In addition, Moonlight predicted FOXM1 as an oncogene with associated amplification in colon adenocarcinoma and lung squamous cell carcinoma43,44. Res. http://bioconductor.org/packages/MoonlightR/, https://gdc.cancer.gov/resources-tcga-users/tcga-code-tables/tcga-study-abbreviations, http://karchinlab.org/data/Protocol/pancan-mutation-set-from-Tokheim-2016.txt.gz, https://gdc.cancer.gov/about-data/publications/mc3-2017, ftp://ftp.sanger.ac.uk/pub/project/cancerrxgene/releases/current_release/sanger1018_brainarray_ensemblgene_rma.txt.gz, ftp://ftp.sanger.ac.uk/pub/project/cancerrxgene/releases/current_release/Cell_Lines_Details.xlsx, Description of Additional Supplementary Files, http://creativecommons.org/licenses/by/4.0/, Experimental analysis of bladder cancer-associated mutations in EP300 identifies EP300-R1627W as a driver mutation, LSM2 is associated with a poor prognosis and promotes cell proliferation, migration, and invasion in skin cutaneous melanoma, COMMD3 loss drives invasive breast cancer growth by modulating copper homeostasis, Unraveling the Drivers of Tumorigenesis in the Context of Evolution: Theoretical Models and Bioinformatics Tools. The forkhead box (Fox) A1 and M1 genes belong to a superfamily of evolutionarily conserved transcriptional factors, and FOXM1 has been shown to be a promising candidate target in the treatment of breast cancer25. a Data used for discovery of oncogenic mediators and controlling mechanisms of cancer driver genes. Med. Histopathology 68, 241253 (2016). Voss, T. C. & Hager, G. L. Dynamic regulation of transcriptional states by chromatin and transcription factors. PubMedGoogle Scholar. In addition, Moonlight identified FLI1 as a tumor suppressor in multiple cancer types, including lung, breast, uterine, and colon (Supplementary Data7). 4a). Subramanian, A. et al. Oncogene 27, 46154624 (2008). Genes were identified as significantly differentially expressed if |logFC|1 and FDR<0.01 in at least one tumor type of the 18 different tumor types, which yielded 13,182 unique genes in total. Nat. Wang, K. et al. Benchmarking mutation effect prediction algorithms using functionally validated cancer-related missense mutations. Key to this effort has been development of computational algorithms to find genes that drive cancer based on their patterns of mutation in large patient cohorts. Experimental validation confirmed 60-85% of predicted mutations as likely drivers. Pan-cancer detection of driver genes at the single-patient resolution In addition, PDGFRA was predicted to be oncogene in thyroid carcinoma and tumor suppressor in colon adenocarcinoma, targeted by 26 (Supplementary Data12). The proto-oncogene c-Kit inhibits tumor growth by behaving as a dependence receptor. A major requirement for drug design is to functionally validate the inhibition potential of targeted cancer driver genes in ex vivo or in vivo cancer models. Moonlight identified mutually exclusive peaks in different regions: open chromatin in the intron region for tumor suppressors (Fig. The steps of FEA involve (i) evaluating if DEGs are involved in a BP through an assessment of the overlap between the list of DEGs and genes relevant to this BP determined by literature mining, and (ii) detecting the BPs mainly enriched by DEGs. To evaluate Moonlights performance, we applied the same ML approach we used for Moonlight to the data used by 20/20+66, and OncodriveRole67 carrying out a leave-one-out cross-validation scheme. Taking a closer look at the tumor suppressors, we found that at least two of these genes, CDKN2A70 and SOCS164, have been linked to colorectal cancer. When we explored the epigenetic modifiers or chromatin accessibility, we observed a global opening of chromatin in the promoter regions for oncogenes predicted by Moonlight. TNBCtype: a subtyping tool for triple-negative breast cancer. Volume 3, Issue 7. Moonlight identified hypermethylated tumor suppressors and hypomethylated oncogenes. We observed that six compounds (methylnorlichexanthone, AG-879, axitinib, ENMD-2076, orantinib, and SU-1498) shared the VEGFR-inhibitor mechanism of action. Hsieh, H. Y. et al. To accomplish this, we performed Pattern Regulation Analysis (Methods), enabling the identification of genes with two distinct patterns. Peer reviewer reports are available. Biochim. We then employed Beegle92 to allow the end user to update the mentioned number of times for BP. Google Scholar. Forbes, S. A. et al. Interestingly, Moonlight also predicted this gene to be a tumor suppressor in bladder urothelial carcinoma with good survival prognosis (log-rank test p=0.022, Fig. Cell 171, 14371452 (2017). As a proof-of-principle, Moonlight accurately predicted cancer driver genes in breast-invasive carcinoma and 17 other cancer types, elucidating their underlying biological mechanisms. While both of these approaches identify cancer driver genes using gene expression data as a major source of information (Fig. The 3p14.2 tumour suppressor ADAMTS9 is inactivated by promoter CpG methylation and inhibits tumour cell growth in breast cancer. Clin. Somatic cells may rapidly acquire mutations, one or two orders of magnitude faster than germline cells [].The majority of these mutations are largely neutral (passenger mutations) in comparison to a few driver mutations that give cells the selective advantage leading to their proliferation []. (Fig. Martinez, J. L. & Baquero, F. Mutation frequencies and antibiotic resistance. Nat. T.C.S., and M.H.B. Ellrott, K. et al. We repeat this procedure 100 times for each of the ten repetitions. Cancer is a complex and heterogeneous disease, hallmarked by the poor regulation of critical functions, such as growth, proliferation, and cell-death pathways. Chromatin accessibility in the promoter region ranges from white (closed) to orange (open). 568 genes identified with the potential to trigger cancer - Medical Xpress Article A complete list of the chosen BPs is reported in Supplementary Data2. Cui, X. et al. Mutations in the KRAS gene are the major driver of pancreatic cancer. In this study, we used a dozen of computational driver gene identification approaches including online resources, offline and online tools to explore most potential breast cancer driver .