--- title: "A workflow to study mechanistic indicators for driver gene prediction with Moonlight" author: "Mona Nourbakhsh^+^ , Astrid Saksager^+^, Nikola Tom, Xi Steven Chen, Antonio Colaprico, Catharina Olsen, Matteo Tiberti, Elena Papaleo" subtitle: ^+^ These authors contributed equally to the paper as first authors. date: "`r Sys.Date()`" output: BiocStyle::html_document: toc: true number_sections: false toc_depth: 2 highlight: haddock vignette: > %\VignetteIndexEntry{A workflow to study mechanistic indicators for driver gene prediction with Moonlight} %\VignetteEngine{knitr::rmarkdown} %\VignetteEncoding{UTF-8} references: - id: ref1 title: Orchestrating high-throughput genomic analysis with Bioconductor author: - family: Huber, Wolfgang and Carey, Vincent J and Gentleman, Robert and Anders, Simon and Carlson, Marc and Carvalho, Benilton S and Bravo, Hector Corrada and Davis, Sean and Gatto, Laurent and Girke, Thomas and others given: journal: Nature methods volume: 12 number: 2 pages: 115-121 issued: year: 2015 - id: ref2 title: GC-content normalization for RNA-Seq data author: - family: Risso, Davide and Schwartz, Katja and Sherlock, Gavin and Dudoit, Sandrine given: journal: BMC bioinformatics volume: 12 number: 1 pages: 480 issued: year: 2011 - id: ref3 title: Evaluation of statistical methods for normalization and differential expression in mRNA-Seq experiments author: - family: Bullard, James H and Purdom, Elizabeth and Hansen, Kasper D and Dudoit, Sandrine given: journal: BMC bioinformatics volume: 11 number: 1 pages: 94 issued: year: 2010 - id: ref4 title: Inferring regulatory element landscapes and transcription factor networks from cancer methylomes author: - family: Yao, L., Shen, H., Laird, P. W., Farnham, P. J., & Berman, B. P. given: journal: Genome biology volume: 16 number: 1 pages: 105 issued: year: 2015 - id: ref5 title: Evaluation of statistical methods for normalization and differential expression in mRNA-Seq experiments author: - family: James H Bullard, Elizabeth Purdom, Kasper D Hansen and Sandrine Dudoit given: journal: BMC Bioinformatics volume: 11 number: 1 pages: 94 issued: year: 2010 - id: ref6 title: GC-content normalization for RNA-Seq data author: - family: Risso, D., Schwartz, K., Sherlock, G., & Dudoit, S. given: journal: BMC Bioinformatics volume: 12 number: 1 pages: 480 issued: year: 2011 - id: ref7 title: Identification of a CpG island methylator phenotype that defines a distinct subgroup of glioma author: - family: Noushmehr, H., Weisenberger, D.J., Diefes, K., Phillips, H.S., Pujara, K., Berman, B.P., Pan, F., Pelloski, C.E., Sulman, E.P., Bhat, K.P. et al. given: journal: Cancer cell volume: 17 number: 5 pages: 510-522 issued: year: 2010 - id: ref8 title: Molecular Profiling Reveals Biologically Discrete Subsets and Pathways of Progression in Diffuse Glioma author: - family: Ceccarelli, Michele and Barthel, Floris P and Malta, Tathiane M and Sabedot, Thais S and Salama, Sofie R and Murray, Bradley A and Morozova, Olena and Newton, Yulia and Radenbaugh, Amie and Pagnotta, Stefano M and others given: journal: Cell URL: "http://doi.org/10.1016/j.cell.2015.12.028" DOI: "10.1016/j.cell.2015.12.028" volume: 164 number: 3 pages: 550-563 issued: year: 2016 - id: ref9 title: Comprehensive molecular profiling of lung adenocarcinoma author: - family: Cancer Genome Atlas Research Network and others given: journal: Nature URL: "http://doi.org/10.1038/nature13385" DOI: "10.1038/nature13385" volume: 511 number: 7511 pages: 543-550 issued: year: 2014 - id: ref10 title: Comprehensive molecular characterization of gastric adenocarcinoma author: - family: Cancer Genome Atlas Research Network and others given: journal: Nature URL: "http://doi.org/10.1038/nature13480" DOI: "10.1038/nature13480" issued: year: 2014 - id: ref11 title: Comprehensive molecular portraits of human breast tumours author: - family: Cancer Genome Atlas Research Network and others given: journal: Nature URL: "http://doi.org/10.1038/nature11412" DOI: "10.1038/nature11412" volume: 490 number: 7418 pages: 61-70 issued: year: 2012 - id: ref12 title: Comprehensive molecular characterization of human colon and rectal cancer author: - family: Cancer Genome Atlas Research Network and others given: journal: Nature URL: "http://doi.org/10.1038/nature11252" DOI: "10.1038/nature11252" volume: 487 number: 7407 pages: 330-337 issued: year: 2012 - id: ref13 title: Genomic classification of cutaneous melanoma author: - family: Cancer Genome Atlas Research Network and others given: journal: Cell URL: "http://doi.org/10.1016/j.cell.2015.05.044" DOI: "10.1016/j.cell.2015.05.044" volume: 161 number: 7 pages: 1681-1696 issued: year: 2015 - id: ref14 title: Comprehensive genomic characterization of head and neck squamous cell carcinomas author: - family: Cancer Genome Atlas Research Network and others given: journal: Nature URL: "http://doi.org/10.1038/nature14129" DOI: "10.1038/nature14129" volume: 517 number: 7536 pages: 576-582 issued: year: 2015 - id: ref15 title: The somatic genomic landscape of chromophobe renal cell carcinoma author: - family: Davis, Caleb F and Ricketts, Christopher J and Wang, Min and Yang, Lixing and Cherniack, Andrew D and Shen, Hui and Buhay, Christian and Kang, Hyojin and Kim, Sang Cheol and Fahey, Catherine C and others given: journal: Cancer Cell URL: "http://doi.org/10.1016/j.ccr.2014.07.014" DOI: "10.1016/j.ccr.2014.07.014" volume: 26 number: 3 pages: 319-330 issued: year: 2014 - id: ref16 title: Comprehensive genomic characterization of squamous cell lung cancers author: - family: Cancer Genome Atlas Research Network and others given: journal: Nature URL: "http://doi.org/10.1038/nature11404" DOI: "10.1038/nature11404" volume: 489 number: 7417 pages: 519-525 issued: year: 2012 - id: ref17 title: Integrated genomic characterization of endometrial carcinoma author: - family: Cancer Genome Atlas Research Network and others given: journal: Nature URL: "http://doi.org/10.1038/nature12113" DOI: "10.1038/nature12113" volume: 497 number: 7447 pages: 67-73 issued: year: 2013 - id: ref18 title: Integrated genomic characterization of papillary thyroid carcinoma author: - family: Cancer Genome Atlas Research Network and others given: journal: Cell URL: "http://doi.org/10.1016/j.cell.2014.09.050" DOI: "10.1016/j.cell.2014.09.050" volume: 159 number: 3 pages: 676-690 issued: year: 2014 - id: ref19 title: The molecular taxonomy of primary prostate cancer author: - family: Cancer Genome Atlas Research Network and others given: journal: Cell URL: "http://doi.org/10.1016/j.cell.2015.10.025" DOI: "10.1016/j.cell.2015.10.025" volume: 163 number: 4 pages: 1011-1025 issued: year: 2015 - id: ref20 title: Comprehensive Molecular Characterization of Papillary Renal-Cell Carcinoma author: - family: Linehan, W Marston and Spellman, Paul T and Ricketts, Christopher J and Creighton, Chad J and Fei, Suzanne S and Davis, Caleb and Wheeler, David A and Murray, Bradley A and Schmidt, Laura and Vocke, Cathy D and others given: journal: NEW ENGLAND JOURNAL OF MEDICINE URL: "http://doi.org/10.1056/NEJMoa1505917" DOI: "10.1056/NEJMoa1505917" volume: 374 number: 2 pages: 135-145 issued: year: 2016 - id: ref21 title: Comprehensive molecular characterization of clear cell renal cell carcinoma author: - family: Cancer Genome Atlas Research Network and others given: journal: Nature URL: "http://doi.org/10.1038/nature12222" DOI: "10.1038/nature12222" volume: 499 number: 7456 pages: 43-49 issued: year: 2013 - id: ref22 title: Comprehensive Pan-Genomic Characterization of Adrenocortical Carcinoma author: - family: Cancer Genome Atlas Research Network and others given: journal: Cancer Cell URL: "http://dx.doi.org/10.1016/j.ccell.2016.04.002" DOI: "10.1016/j.ccell.2016.04.002" volume: 29 pages: 43-49 issued: year: 2016 - id: ref23 title: Complex heatmaps reveal patterns and correlations in multidimensional genomic data author: - family: Gu, Zuguang and Eils, Roland and Schlesner, Matthias given: journal: Bioinformatics URL: "http://dx.doi.org/10.1016/j.ccell.2016.04.002" DOI: "10.1016/j.ccell.2016.04.002" pages: "btw313" issued: year: 2016 - id: ref24 title: "TCGA Workflow: Analyze cancer genomics and epigenomics data using Bioconductor packages" author: - family: Silva, TC and Colaprico, A and Olsen, C and D'Angelo, F and Bontempi, G and Ceccarelli, M and Noushmehr, H given: journal: F1000Research URL: "http://dx.doi.org/10.12688/f1000research.8923.1" DOI: "10.12688/f1000research.8923.1" volume: 5 number: 1542 issued: year: 2016 - id: ref25 title: "TCGAbiolinks: an R/Bioconductor package for integrative analysis of TCGA data" author: - family: Colaprico, Antonio and Silva, Tiago C. and Olsen, Catharina and Garofano, Luciano and Cava, Claudia and Garolini, Davide and Sabedot, Thais S. and Malta, Tathiane M. and Pagnotta, Stefano M. and Castiglioni, Isabella and Ceccarelli, Michele and Bontempi, Gianluca and Noushmehr, Houtan given: journal: Nucleic Acids Research URL: "http://dx.doi.org/10.1093/nar/gkv1507" DOI: "10.1093/nar/gkv1507" volume: 44 number: 8 pages: e71 issued: year: 2016 - id: ref26 title: "Interpreting pathways to discover cancer driver genes with Moonlight" author: - family: Colaprico, Antonio and Olsen, Catharina and Bailey, Matthew H. and Odom, Gabriel J. and Terkelsen, Thilde and Silva, Tiago C. and Olsen, André V. and Cantini, Laura and Zinovyev, Andrei and Barillot, Emmanuel and Noushmehr, Houtan and Bertoli, Gloria and Castiglioni, Isabella and Cava, Claudia and Bontempi, Gianluca and Chen, Xi Steven and Papaleo, Elena given: journal: Nature Communications URL: "https://doi.org/10.1038/s41467-019-13803-0" DOI: "10.1038/s41467-019-13803-0" volume: 11 number: 69 issued: year: 2020 - id: ref27 title: "A workflow to study mechanistic indicators for driver gene prediction with Moonlight" author: - family: Nourbakhsh, Mona and Saksager, Astrid and Tom, Nikola and Chen, Xi Steven and Colaprico, Antonio and Olsen, Catharina and Tiberti, Matteo and Papaleo, Elena given: journal: Briefings in Bioinformatics URL: "https://doi.org/10.1093/bib/bbad274" DOI: "10.1093/bib/bbad274" issued: year: 2023 - id: ref28 title: "EpiMix is an integrative tool for epigenomic subtyping using DNA methylation" author: - family: Zheng, Yuanning and Jun, John and Brennan, Kevin and Gevaert, Olivier given: journal: Cell Reports Methods URL: "https://doi.org/10.1016/j.crmeth.2023.100515" DOI: "10.1016/j.crmeth.2023.100515" issued: year: 2023 - id: ref29 title: "CScape-somatic: distinguishing driver and passenger point mutations in the cancer genome" author: - family: Rogers, Mark F and Gaunt, Tom R and Campbell, Colin given: journal: Bioinformatics URL: "https://doi.org/10.1093/bioinformatics/btaa242" DOI: "10.1093/bioinformatics/btaa242" issued: year: 2020 --- ```{r setup, include=FALSE} knitr::opts_chunk$set(dpi = 72) knitr::opts_chunk$set(cache=FALSE) ``` # Abstract In order to make light of cancer development, it is crucial to understand which genes play a role in the mechanisms linked to this disease and moreover which role that is. Commonly biological processes such as proliferation and apoptosis have been linked to cancer progression. We have developed the Moonlight framework that allows for prediction of cancer driver genes through multi-omics data integration. Based on expression data we perform functional enrichment analysis, infer gene regulatory networks and upstream regulator analysis to score the importance of well-known biological processes with respect to the studied cancer. We then use these scores to predict oncogenic mediators with two specific roles: genes that potentially act as tumor suppressor genes (TSGs) and genes that potentially act as oncogenes (OCGs). This constitutes Moonlight's primary layer. As gene expression data alone does not explain the cancer phenotypes, a second layer of evidence is needed. We have integrated three secondary layers, one based on mutations, one based on abnormal DNA methylation patterns, for prediction of the driver genes, and one baed on the effect of mutations on transcription factors. The mutational layer predicts driver mutations in the oncogenic mediators and thereby allows for the prediction of cancer driver genes using the driver mutation prediction tool CScape-somatic. The methylation layer investigates abnormal DNA methylation patterns and differentially methylated CpG sites in the oncogenic mediators using the tool EpiMix and uses patterns of hypo- and hypermethylation for prediction of the driver genes. The transcription factor layer checks which transcription factors are important for determining the transcription of possible driver genes, and identifies mutations on such transcription factors from a cancer mutation dataset that are able to destabilize them and therefore likely affect transcription. Overall, this methodology not only allows us to identify genes with dual role (TSG in one cancer type and OCG in another) but also to elucidate the underlying biological processes. # Introduction Cancer development is influenced by (epi)genetic alterations in two distinctly different categories of genes, known as tumor suppressor genes (TSG) and oncogenes (OCG). The occurrence of these alterations in genes of the first category leads to faster cell proliferation while alterations in genes of second category increases or changes their function. In 2020, we developed the Moonlight framework that allows for prediction of cancer driver genes [@ref26]. Here, gene expression data are integrated together with biological processes and gene regulatory networks to score the importance of well-known biological processes with respect to the studied cancer. These scores are used to predict oncogenic mediators: putative TSGs and putative OCGs. As gene expression data alone is not enough to explain the deregulation of the genes, a second layer of evidence is needed. For this reason, we integrated a secondary mutational layer which predicts driver mutations and passenger mutations in the oncogenic mediators. The prediction of the driver mutations are carried out using the CScape-somatic [@ref29] driver mutation prediction tool. Those oncogenic mediators with at least one driver mutation are retained as the final set of driver genes [@ref27]. Moreover, we have integrated a secondary methylation layer that investigates abnormal DNA methylation patterns in the oncogenic mediators using the tool EpiMix [@ref28]. Those oncogenic mediators demonstrating hypo- and hypermethylation patterns are predicted as OCGs and TSGs, respectively. These new functionalities are released in the updated version of Moonlight to produce Moonlight2R. A third transcription factor layer checks which transcription factors are important for determining the transcription of possible driver genes, and if any mutation available in a cancer mutation dataset is able to destabilize them and therefore likely affect transcription. The mutations can be provided as MAF-formatted files, while predictions on whether any mutation on a transcription factor can be destabilizing are available as the [MAVISp database](https://osf.io/ufpzm/). MAVISp is our computational framework for the effect of predictions of mutations using structural methods; you can read more about it on our [preprint](https://www.biorxiv.org/content/10.1101/2022.10.22.513328v5). # Moonlight's pipeline Moonlight's pipeline is shown below: ```{r, fig.width=3, echo = FALSE, fig.align="center",hide=TRUE, message=FALSE,warning=FALSE} knitr::include_graphics("Moonlight2_pipeline_upd.png") ``` # Moonlight's proposed workflow The proposed pipeline consists of following eight steps: 1. The input to Moonlight is a set of differentially expressed genes between two biological conditions such as cancer and healthy samples or two cancer subtyoes. Besides differentially expressed genes, gene expression data is also needed. Mutation data and/or methylation data are also needed. 2. **FEA** Functional Enrichment Analysis, using Fisher's test, to identify gene sets (with biological functions linked to cancer) significantly enriched by regulated genes (RG). 3. **GRN** Gene Regulatory Network inferred between each single DEG (sDEG) and all genes by means of mutual information, obtaining for each DEG a list of RG. 4. **URA** Upstream Regulator Analysis for DEGs in each enriched gene set, we applied z-score being the ratio between the sum of all predicted effects for all the gene involved in the specific function and the square-root of the number of all genes. 5. **PRA** Pattern Recognition Analysis identifies putative TSGs (down) and OCGs (up). We either use user defined biological processes or random forests. 6. **DMA** Driver Mutation Analysis identifies driver mutations in the putative TSGs and OCGs through the driver mutation prediction tool, CScape-somatic. 7. **GMA** Gene Methylation Analysis identifies abnormal DNA methylation patterns in the putative TSGs and OCGs through the tool, EpiMix. 8. We apply the above procedure to multiple cancer types to obtain cancer-specific lists of TSGs and OCGs. We compare the lists for each cancer: if a sDEG is TSG in a cancer and OCG in another we define it as dual-role TSG-OCG. Otherwise if we find a sDEG defined as OCG or TSG only in one tissue we define it as tissue specific biomarker. 9. We use the COSMIC database to define a list of gold standard TSG and OCGs to assess the accuracy of the proposed method. # Installation To install Moonlight2R use the code below. ## Installation from BioConductor ```{r, eval = FALSE} if (!require("BiocManager", quietly = TRUE)) install.packages("BiocManager") BiocManager::install("Moonlight2R") ``` ## Installation from GitHub First, install `devtools` or if you already have it installed, load it. ```{r, eval = FALSE} install.packages("devtools") library(devtools) ``` Install Moonlight2R from GitHub: ```{r, eval = FALSE} devtools::install_github(repo = "ELELAB/Moonlight2R") ``` ## Installation from GitHub with accompanying vignette First, install the BiocStyle Bioconductor package. ```{r, eval = FALSE} if (!require("BiocManager", quietly = TRUE)) install.packages("BiocManager") BiocManager::install("BiocStyle") ``` Then install Moonlight2R with its accompanying vignette. ```{r, eval = FALSE} devtools::install_github(repo = "ELELAB/Moonlight2R", build_vignettes = TRUE) ``` You can view the vignette in the following way. ```{r, eval = FALSE} vignette( "Moonlight2R", package="Moonlight2R") ``` # Load libraries ```{r, eval = TRUE} library(Moonlight2R) library(magrittr) library(dplyr) ``` # `Obtain Input` The input to Moonlight is a set of differentially expressed genes and gene expression data. Mutation and/or methylation data are also needed. These data should be available for the same sample sets. Gene expression data, mutation data, methylation data and differentially expressed genes can for example be obtained from TCGA using the R package TCGAbiolinks. Help documents on how to use TCGAbiolinks are available [here](https://bioconductor.org/packages/release/bioc/vignettes/TCGAbiolinks/inst/doc/index.html). To find other examples of usage of TCGAbiolinks on TCGA cancer types see our [GitHub repository](https://github.com/ELELAB/LUAD_LUSC_TCGA_comparison). Example data of the input (differentially expressed genes, gene expression data, mutation data, and methylation data) are stored in the Moonlight2R package: ```{r, eval = TRUE} data(DEGsmatrix) data(dataFilt) data(dataMAF) data(GEO_TCGAtab) data(LOC_transcription) data(LOC_translation) data(LOC_protein) data(Oncogenic_mediators_mutation_summary) data(DEG_Mutations_Annotations) data(dataMethyl) data(DEG_Methylation_Annotations) data(Oncogenic_mediators_methylation_summary) data(MetEvidenceDriver) data(LUAD_sample_anno) ``` # `Download`: Get GEO data You can search GEO data using the `getDataGEO` function. GEO_TCGAtab: a 18x12 matrix that provides the GEO data set we matched to one of the 18 given TCGA cancer types ```{r, eval = TRUE, echo = TRUE} knitr::kable(GEO_TCGAtab, digits = 2, caption = "Table with GEO data set matched to one of the 18 given TCGA cancer types ", row.names = TRUE) ``` ## `getDataGEO`: Search by cancer type and data type [Gene Expression] The user can query and download the cancer types supported by GEO, using the function `getDataGEO`: ```{r, eval = TRUE, echo = TRUE, results='hide', warning = FALSE, message = FALSE} dataFilt_GEO <- getDataGEO(GEOobject = "GSE20347", platform = "GPL571") ``` ```{r, eval = TRUE, echo = TRUE, results='hide', warning = FALSE, message = FALSE} dataFilt_GEO <- getDataGEO(TCGAtumor = "ESCA") ``` ## `FEA`: Functional Enrichment Analysis The user can perform a functional enrichment analysis using the function `FEA`. For each DEG in the gene set a z-score is calculated. This score indicates how the genes act in the gene set. ```{r, eval = TRUE, echo = TRUE, results='hide'} data(DEGsmatrix) data(DiseaseList) data(EAGenes) dataFEA <- FEA(DEGsmatrix = DEGsmatrix) ``` The output can be visualized with a FEA plot. ## `FEAplot`: Functional Enrichment Analysis Plot The user can plot the result of a functional enrichment analysis using the function `plotFEA`. A negative z-score indicates that the process' activity is decreased. A positive z-score indicates that the process' activity is increased. ```{r, eval = TRUE, echo = TRUE, message=FALSE, results='hide', warning=FALSE} plotFEA(dataFEA = dataFEA, additionalFilename = "_exampleVignette", height = 10, width = 20) ``` The figure generated by the above code is shown below: ```{r, fig.width=3, echo = FALSE, fig.align="center",hide=TRUE, message=FALSE,warning=FALSE} knitr::include_graphics("FEAplot.gif") ``` ## `GRN`: Gene Regulatory Network The user can perform a gene regulatory network analysis using the function `GRN` which infers the network using the parmigene package. For illustrative purposes and to decrease runtime, we have set `nGenesPerm = 5` and `nBoot = 5` in the example below, however, we recommend setting these parameters to `nGenesPerm = 2000` and `nBoot = 400` to achieve optimal results, as they are set by default in the function arguments. ```{r, eval = TRUE} data(DEGsmatrix) data(dataFilt) dataGRN <- GRN(DEGsmatrix = DEGsmatrix, TFs = sample(rownames(DEGsmatrix), 100), normCounts = dataFilt, nGenesPerm = 5, kNearest = 3, nBoot = 5, DiffGenes = TRUE) ``` ## `URA`: Upstream Regulator Analysis The user can perform upstream regulator analysis using the function `URA`. This function is applied to each DEG in the enriched gene set and its neighbors in the GRN. ```{r, eval = TRUE, echo = TRUE, results='hide'} data(dataGRN) data(DEGsmatrix) data(DiseaseList) data(EAGenes) dataFEA <- FEA(DEGsmatrix = DEGsmatrix) BPselected <- dataFEA$Diseases.or.Functions.Annotation[1:5] dataURA <- URA(dataGRN = dataGRN, DEGsmatrix = DEGsmatrix, BPname = BPselected, nCores=1) ``` ## `PRA`: Pattern Regognition Analysis The user can retrieve putative TSG/OCG using either selected biological processes in the expert-based approach or a random forest classifier trained on known COSMIC OCGs/TSGs in the machine learning approach. The user must enter the chosen biological processes in the `BPname` argument to use the expert-based approach or set `BPname = NULL` to use the machine learning approach. ```{r, eval = TRUE} data(dataURA) data(tabGrowBlock) data(knownDriverGenes) dataPRA <- PRA(dataURA = dataURA, BPname = c("apoptosis","proliferation of cells"), thres.role = 0) ``` ## `DMA`: Driver Mutation Analysis The user can identify driver mutations with `DMA` in the oncogenic mediators established by `PRA`. The passenger or driver status is estimated with CScape-somatic [@ref29]. This function will further generate three files: DEG_Mutations_Annotations.rda, Oncogenic_mediators_mutation_summary.rda and cscape_somatic_output.rda. These will be placed in the specified results-folder. The user needs to download two CScape-somatic files in order to run DMA named css_coding.vcf.gz and css_noncoding.vcf.gz, respectively. These two files can be downloaded at http://cscape-somatic.biocompute.org.uk/#download. The corresponding .tbi files (css_coding.vcf.gz.tbi and css_noncoding.vcf.gz.tbi) must also be downloaded and be placed in the same folder. ```{r, eval = FALSE} data(dataPRA) data(dataMAF) data(DEGsmatrix) data(LOC_transcription) data(LOC_translation) data(LOC_protein) data(NCG) data(EncodePromoters) dataDMA <- DMA(dataMAF = dataMAF, dataDEGs = DEGsmatrix, dataPRA = dataPRA, results_folder = "DMAresults", coding_file = "css_coding.vcf.gz", noncoding_file = "css_noncoding.vcf.gz") ``` ## `GMA`: Gene Methylation Analysis The user can predict driver genes with `GMA` following prediction of the oncogenic mediators established by `PRA`. `GMA` predicts driver genes by investigating abnormal DNA methylation patterns in the oncogenic mediators using the tool EpiMix [@ref28]. Oncogenic mediators with hypo- and hypermethylation patterns are predicted as OCGs and TSGs, respectively. This function will generate these files: DEG_Methylation_Annotations.rda, Oncogenic_mediators_methylation_summary.rda, EpiMix_Results_Enhancer.rds, EpiMix_Results_Regular.rds, FunctionalPairs_Enhancer.csv, FunctionalPairs_Regular.csv, and FunctionalProbes_Regular.rds. These will be placed in the specified results-folder. ```{r, eval = TRUE} data("dataMethyl") data("dataFilt") data("dataPRA") data("DEGsmatrix") data("LUAD_sample_anno") data("NCG") data("EncodePromoters") data("MetEvidenceDriver") # Subset column names (sample names) in expression data to patient level pattern <- "^(.{4}-.{2}-.{4}-.{2}).*" colnames(dataFilt) <- sub(pattern, "\\1", colnames(dataFilt)) dataGMA <- GMA(dataMET = dataMethyl, dataEXP = dataFilt, dataPRA = dataPRA, dataDEGs = DEGsmatrix, sample_info = LUAD_sample_anno, met_platform = "HM450", prevalence_filter = NULL, output_dir = "./GMAresults", cores = 1, roadmap.epigenome.ids = "E096", roadmap.epigenome.groups = NULL) ``` ## `GLS`: Gene Literature Search The user can perform a literature search on driver genes using the `GLS` function. This function takes as input driver genes, a query and maximum number of records to retrieve from PubMed. Standard PubMed syntax can be used in the query. For example, Boolean operators AND, OR, NOT can be applied and tags such as [AU], [TITLE/ABSTRACT], [Affiliation] can be used. `GLS` fetches data of PubMed records matching the specified query and outputs PubMed IDs matching the query along with doi, title, abstract, year of publication, mesh terms, and total number of PubMed publications. This is done for each of the genes supplied in the input. ```{r, eval = TRUE} genes_query <- "BRCA1" dataGLS <- GLS(genes = genes_query, query_string = "AND cancer AND driver") head(dataGLS) ``` ## Transcription factor-based layer of secondary evidence We now introduce the third layer of secondary evidence. It is based on the assumption that stability changes due to mutations on transcription factors can influence transcription. Here, we use data on change of stability upon mutation available in the MAVISp database, as well as data from the [TRRUST database](https://www.grnpedia.org/trrust/) and MAF files from TCGA to i) identify transcription factors that have an effect on the transcription of genes of interest ii) identify potentially damaging mutations on them from MAF files and iii) use MAVISp data to assess the affect of such mutations, whether they might destabilizing or not. We release the TRRUST-derived data within the Moonlight2R package with a Creative Commons Attribution-ShareAlike 4.0 International license. ## `loadMAVISp`: Helper function for loading the MAVISp database The first step to perform this is loading the data form the MAVISp database. Notice that we have included a very small portion of the MAVISp database with Moonlight2R for testing purposes; we recommend using the full database available at the [MAVISp OSF repository](https://osf.io/ufpzm/). Also see the [MAVISp preprint](https://www.biorxiv.org/content/10.1101/2022.10.22.513328v5) for specifics on what datasets are available in MAVISp. The `loadMAVISp` function is used to load the files available from the MAVISp database. The user will need to decide on whether they want to import the simple or ensemble mode and in the case of ensemble mode, choose the ensemble they want to import. Furthermore, the user can specify specific proteins to load. The output is a list of tibbles, each one correponding to a file in the MAVISp database. The output is used directly in the `TFinfluence` function. ```{r, eval = TRUE} mavisp_db_location <- system.file('extdata', 'mavisp_db', package='Moonlight2R') specific_protein <- loadMAVISp(mavispDB = mavisp_db_location, mode = 'simple', proteins_of_interest = c('RUNX1')) all_proteins <- loadMAVISp(mavispDB = mavisp_db_location, mode = 'simple') ensemble <- loadMAVISp(mavispDB = mavisp_db_location, mode = 'ensemble', ensemble = 'cabsflex') ``` ## `TFinfluence`: Effect of mutations on transcription factors As previosuly introduceed, the `TFinfluence` function can be used as after the PRA function, to identify if there are mutations in the transcription factors of the DEGs. The function uses data from TRRUST to couple the DEGs with their transcription factor (TF). The mutations in the TFs come from a MAF file, and the effect of the mutation is obtained from the MAVISp database, available at an [OSF repository](https://osf.io/ufpzm/). ```{r, eval = TRUE} data(dataPRA) data(DEGsmatrix) data(dataTRRUST) data(dataMAF) data(dataMAVISp) TFresults <- TFinfluence(dataTRRUST = dataTRRUST, dataMAF = dataMAF, dataDEGs = DEGsmatrix, dataPRA = dataPRA, dataMAVISp = dataMAVISp, stabClassMAVISp = 'rasp') ``` ## `Level of consequence`: Effect of mutations on three different levels The user can investigate the predicted effect of different mutation types on the transcriptional level through the table LOC_transcription: ```{r} knitr::kable(LOC_transcription) ``` The user can investigate the predicted effect of different mutation types on the translational level through the table LOC_translation: ```{r} knitr::kable(LOC_translation) ``` The user can investigate the predicted effect of different mutation types on the protein level through the table LOC_protein: ```{r} knitr::kable(LOC_protein) ``` ## `plotNetworkHive`: GRN hive visualization taking into account COSMIC cancer genes In the following plot the nodes are separated into three groups: known tumor suppressor genes (yellow), known oncogenes (green) and the rest (gray). ```{r, eval = TRUE, echo = TRUE, results='hide', warning = FALSE, message = FALSE} data(knownDriverGenes) data(dataGRN) plotNetworkHive(dataGRN, knownDriverGenes, 0.55) ``` ## `plotDMA`: Heatmap of the driver/passenger status of mutations in TSGs/OCGs In the following plot the driver genes with driver mutations are shown. ```{r, eval = TRUE, warning = FALSE, message = FALSE, include=TRUE} data(dataDMA) data(DEG_Mutations_Annotations) data(Oncogenic_mediators_mutation_summary) plotDMA(DEG_Mutations_Annotations, Oncogenic_mediators_mutation_summary, type = 'complete', additionalFilename = "") ``` ```{r, fig.width=3, fig.height=4, echo = FALSE, fig.align="center",hide=TRUE, message=FALSE,warning=FALSE} knitr::include_graphics("heatmap_complete.gif") ``` ## `plotMoonlight`: Heatmap of Moonlight Gene Z-scores for mutation-driven TSGs/OCGs In the following plot the top 50 genes with the most driver mutations are visualised. The values are the moonlight gene z-score for the two biological processes ```{r, eval = TRUE, echo = TRUE, results='hide', warning = FALSE, message = FALSE} data(DEG_Mutations_Annotations) data(Oncogenic_mediators_mutation_summary) data(dataURA_plot) plotMoonlight(DEG_Mutations_Annotations, Oncogenic_mediators_mutation_summary, dataURA_plot, gene_type = "drivers", n = 50) ``` ```{r, fig.width=3, echo = FALSE, fig.align="center",hide=TRUE, message=FALSE,warning=FALSE} knitr::include_graphics("moonlight_heatmap.gif") ``` ## `plotGMA`: Heatmap of hypo/hyper/dual methylated CpG sites in TSGs/OCGs This function plots results of the `GMA`. It visualizes the number of hypo/hyper/dual methylated CpG sites in oncogenic mediators or in a user-supplied gene list. The results are visualized either in a single heatmap or split into different ones which is specified in the function's three modes: split, complete and genelist. ```{r, eval = TRUE, warning = FALSE, message = FALSE, include = TRUE} data("DEG_Methylation_Annotations") data("Oncogenic_mediators_methylation_summary") genes <- c("ACAN", "ACE2", "ADAM19", "AFAP1L1") plotGMA(DEG_Methylation_Annotations = DEG_Methylation_Annotations, Oncogenic_mediators_methylation_summary = Oncogenic_mediators_methylation_summary, type = "genelist", genelist = genes, additionalFilename = "./GMAresults/") ``` ```{r, fig.width=3, fig.height=4, echo = FALSE, fig.align="center",hide=TRUE, message=FALSE,warning=FALSE} knitr::include_graphics("heatmap_genelist_met.gif") ``` ## `plotMoonlightMet`: Heatmap of Moonlight Gene Z-scores for methylation-driven TSGs/OCGs This function visualizes the effect of genes on biological processes and total number of hypo/hyper/dual methylated CpG sites in genes. ```{r, eval = TRUE, warning = FALSE, message = FALSE, include = TRUE} data("DEG_Methylation_Annotations") data("Oncogenic_mediators_methylation_summary") data("dataURA_plot") genes <- c("ACAN", "ACE2", "ADAM19", "AFAP1L1") plotMoonlightMet(DEG_Methylation_Annotations = DEG_Methylation_Annotations, Oncogenic_mediators_methylation_summary = Oncogenic_mediators_methylation_summary, dataURA = dataURA_plot, genes = genes, additionalFilename = "./GMAresults/") ``` ```{r, fig.width=3, fig.height=4, echo = FALSE, fig.align="center",hide=TRUE, message=FALSE,warning=FALSE} knitr::include_graphics("moonlight_heatmap_met.gif") ``` ## `plotMetExp`: Visualize results from EpiMix of expression and methylation in genes This function visualizes results of EpiMix [@ref28] in three plots: one that visualizes the distribution of DNA methylation data (MixtureModelPlot), one that visualizes gene expression levels (ViolinPlot) and one that visualizes relationship between DNA methylation and gene expression (CorrelationPlot). ```{r, eval = TRUE, warning = FALSE, message = FALSE, include = TRUE} data("EpiMix_Results_Regular") data("dataMethyl") data("dataFilt") # Subset column names (sample names) in expression data to patient level pattern <- "^(.{4}-.{2}-.{4}-.{2}).*" colnames(dataFilt) <- sub(pattern, "\\1", colnames(dataFilt)) plotMetExp(EpiMix_results = EpiMix_Results_Regular, probe_name = "cg03035704", dataMET = dataMethyl, dataEXP = dataFilt, gene_of_interest = "ACVRL1", additionalFilename = "./GMAresults/") ``` ```{r, fig.width=3, fig.height=4, echo = FALSE, fig.align="center",hide=TRUE, message=FALSE,warning=FALSE} knitr::include_graphics("plotMetExp.png") ``` # Moonlight Analysis: Case Studies ### Introduction This vignette shows a complete workflow of the 'Moonlight2R' package. The code is divided into the following case studies: * 1. Predicting oncogenic mediators using Moonlight's primary layer * 2. Moonlight pipeline in one function * 3. Moonlight with driver mutation analysis * 4. Moonlight with gene methylation analysis ## Case study n. 1: Predicting oncogenic mediators using Moonlight's primary layer For illustrative purposes and to decrease runtime, we have set `nGenesPerm = 5` and `nBoot = 5` in the call of `GRN` in the following code block, however, we recommend setting these parameters to `nGenesPerm = 2000` and `nBoot = 400` to achieve optimal results, as they are set by default in the function arguments. ```{r,eval = TRUE,echo=TRUE,message=FALSE,warning=FALSE, results='hide'} data(DEGsmatrix) data(dataFilt) data(DiseaseList) data(EAGenes) data(tabGrowBlock) data(knownDriverGenes) dataFEA <- FEA(DEGsmatrix = DEGsmatrix) dataGRN <- GRN(TFs = sample(rownames(DEGsmatrix), 100), DEGsmatrix = DEGsmatrix, DiffGenes = TRUE, normCounts = dataFilt, nGenesPerm = 5, nBoot = 5, kNearest = 3) dataURA <- URA(dataGRN = dataGRN, DEGsmatrix = DEGsmatrix, BPname = c("apoptosis", "proliferation of cells")) dataDual <- PRA(dataURA = dataURA, BPname = c("apoptosis", "proliferation of cells"), thres.role = 0) oncogenic_mediators <- list("TSG"=names(dataDual$TSG), "OCG"=names(dataDual$OCG)) ``` ## `plotURA`: Upstream regulatory analysis plot The user can plot the result of the upstream regulatory analysis using the function `plotURA`. ```{r, eval = TRUE,message=FALSE,warning=FALSE, results='hide'} data(dataURA) plotURA(dataURA = dataURA, additionalFilename = "_exampleVignette") ``` The figure resulted from the code above is shown below: ```{r, fig.width=3, echo = FALSE, fig.align="center",hide=TRUE, message=FALSE,warning=FALSE} knitr::include_graphics("URAplot.gif") ``` ## Case study n. 2: Moonlight pipeline in one function For illustrative purposes and to decrease runtime, we have set `nGenesPerm = 5` and `nBoot = 5` in the following code block, however, we recommend setting these parameters to `nGenesPerm = 2000` and `nBoot = 400` to achieve optimal results, as they are set by default in the function arguments. ```{r,eval = FALSE,echo=TRUE,message=FALSE,warning=FALSE} data(dataFilt) data(DEGsmatrix) data(dataMAF) data(DiseaseList) data(EAGenes) data(tabGrowBlock) data(knownDriverGenes) data(LOC_transcription) data(LOC_translation) data(LOC_protein) data(NCG) data(EncodePromoters) listMoonlight <- moonlight(dataDEGs = DEGsmatrix, dataFilt = dataFilt, nTF = 100, DiffGenes = TRUE, nGenesPerm = 5, nBoot = 5, BPname = c("apoptosis","proliferation of cells"), dataMAF = dataMAF, path_cscape_coding = "css_coding.vcf.gz", path_cscape_noncoding = "css_noncoding.vcf.gz") save(listMoonlight, file = paste0("listMoonlight_ncancer4.Rdata")) ``` ## `plotCircos`: Moonlight Circos Plot An example of running Moonlight on five cancer types is visualized below in a circos plot. Outer ring: color by cancer type, Inner ring: OCGs and TSGs, Inner connections: green: common OCGs yellow: common TSGs red: possible dual role ```{r, eval = TRUE, echo = TRUE, results='hide', warning = FALSE, message = FALSE} data(listMoonlight) plotCircos(listMoonlight = listMoonlight, additionalFilename = "_ncancer5") ``` The figure generated by the code above is shown below: ```{r, fig.width=3, echo = FALSE, fig.align="center",hide=TRUE, message=FALSE,warning=FALSE} knitr::include_graphics("circos_ocg_tsg_ncancer5.gif") ``` ## Case study n. 3: Moonlight with driver mutation analysis For illustrative purposes and to decrease runtime, we have set `nGenesPerm = 5` and `nBoot = 5` in the call of `GRN` in the following code block, however, we recommend setting these parameters to `nGenesPerm = 2000` and `nBoot = 400` to achieve optimal results, as they are set by default in the function arguments. ```{r,eval = FALSE,echo=TRUE,message=FALSE,warning=FALSE} data(DEGsmatrix) data(dataFilt) data(dataMAF) data(DiseaseList) data(EAGenes) data(tabGrowBlock) data(knownDriverGenes) data(LOC_transcription) data(LOC_translation) data(LOC_protein) data(NCG) data(EncodePromoters) # Perform gene regulatory network analysis dataGRN <- GRN(TFs = rownames(DEGsmatrix), DEGsmatrix = DEGsmatrix, DiffGenes = TRUE, normCounts = dataFilt, nGenesPerm = 5, kNearest = 3, nBoot = 5) # Perform upstream regulatory analysis # As example, we use apoptosis and proliferation of cells as the biological processes dataURA <- URA(dataGRN = dataGRN, DEGsmatrix = DEGsmatrix, BPname = c("apoptosis", "proliferation of cells"), nCores = 1) # Perform pattern recognition analysis dataPRA <- PRA(dataURA = dataURA, BPname = c("apoptosis", "proliferation of cells"), thres.role = 0) # Perform driver mutation analysis dataDMA <- DMA(dataMAF = dataMAF, dataDEGs = DEGsmatrix, dataPRA = dataPRA, results_folder = "DMAresults", coding_file = "css_coding.vcf.gz", noncoding_file = "css_noncoding.vcf.gz") ``` Next, we analyze the predicted driver genes and their mutations. ```{r, eval = TRUE} data(Oncogenic_mediators_mutation_summary) data(DEG_Mutations_Annotations) # Extract oncogenic mediators that contain at least one driver mutation # These are the driver genes knitr::kable(Oncogenic_mediators_mutation_summary %>% filter(!is.na(CScape_Driver))) # Extract mutation annotations of the predicted driver genes driver_mut <- DEG_Mutations_Annotations %>% filter(!is.na(Moonlight_Oncogenic_Mediator), CScape_Mut_Class == "Driver") # Extract driver genes with a predicted effect on the transcriptional level transcription_mut <- Oncogenic_mediators_mutation_summary %>% filter(!is.na(CScape_Driver)) %>% filter(Transcription_mut_sum > 0) # Extract mutation annotations of predicted driver genes that have a driver mutation # in its promoter region with a potential effect on the transcriptional level promoters <- DEG_Mutations_Annotations %>% filter(!is.na(Moonlight_Oncogenic_Mediator), CScape_Mut_Class == "Driver", Potential_Effect_on_Transcription == 1, Annotation == 'Promoter') ``` ## Case study n. 4: Moonlight with gene methylation analysis For illustrative purposes and to decrease runtime, we have set `nGenesPerm = 5` and `nBoot = 5` in the call of `GRN` in the following code block, however, we recommend setting these parameters to `nGenesPerm = 2000` and `nBoot = 400` to achieve optimal results, as they are set by default in the function arguments. ```{r,eval = TRUE,echo=TRUE,message=FALSE,warning=FALSE,results='hide'} data(DEGsmatrix) data(dataFilt) data(dataMAF) data(DiseaseList) data(EAGenes) data(tabGrowBlock) data(knownDriverGenes) data(LOC_transcription) data(LOC_translation) data(LOC_protein) data(NCG) data(EncodePromoters) data(dataMethyl) data(LUAD_sample_anno) data(MetEvidenceDriver) # Perform gene regulatory network analysis dataGRN <- GRN(TFs = rownames(DEGsmatrix), DEGsmatrix = DEGsmatrix, DiffGenes = TRUE, normCounts = dataFilt, nGenesPerm = 5, kNearest = 3, nBoot = 5) # Perform upstream regulatory analysis # As example, we use apoptosis and proliferation of cells as the biological processes dataURA <- URA(dataGRN = dataGRN, DEGsmatrix = DEGsmatrix, BPname = c("apoptosis", "proliferation of cells"), nCores = 1) # Perform pattern recognition analysis dataPRA <- PRA(dataURA = dataURA, BPname = c("apoptosis", "proliferation of cells"), thres.role = 0) # Subset column names (sample names) in expression data to patient level pattern <- "^(.{4}-.{2}-.{4}-.{2}).*" colnames(dataFilt) <- sub(pattern, "\\1", colnames(dataFilt)) # Perform gene methylation analysis dataGMA <- GMA(dataMET = dataMethyl, dataEXP = dataFilt, dataPRA = dataPRA, dataDEGs = DEGsmatrix, sample_info = LUAD_sample_anno, met_platform = "HM450", prevalence_filter = NULL, output_dir = "./GMAresults", cores = 1, roadmap.epigenome.ids = "E096", roadmap.epigenome.groups = NULL) ``` ## Citation Please cite the MoonlightR and Moonlight2R packages: * "Interpreting pathways to discover cancer driver genes with Moonlight." Nature Communications (2020): [10.1038/s41467-019-13803-0](https://doi.org/10.1038/s41467-019-13803-0). [@ref26] * "A workflow to study mechanistic indicators for driver gene prediction with Moonlight." Briefings in Bioinformatics (2023): [10.1093/bib/bbad274](https://doi.org/10.1093/bib/bbad274). [@ref27] Related publications to this vignette: * "TCGAbiolinks: an R/Bioconductor package for integrative analysis of TCGA data." Nucleic acids research (2015): [gkv1507](http://dx.doi.org/doi:10.1093/nar/gkv1507). [@ref25] * "TCGA Workflow: Analyze cancer genomics and epigenomics data using Bioconductor packages". F1000Research [10.12688/f1000research.8923.1](http://dx.doi.org/doi:10.12688/f1000research.8923.1) [@ref24] ****** Session Information ****** ```{r sessionInfo} sessionInfo() ``` # References