DEGenR: An R Shiny app for differential gene expression and enrichment analysis
Contributors
Project leader:
Project manager:
Project member:
- 1. UC San Diego
Description
Abstract
This is the first public release of the DEGenR, an interactive Shiny app that provides integrated tools for performing differential gene expression, ranked-based ontological geneset and pathway enrichment analysis, and transcription factor regulatory analysis from user-uploaded RNA-seq counts or microarray and RNA-seq datasets available at the Gene Expression Omnibus (GEO) and recount2.
Introduction
Differential gene expression testing followed by pathway and transcription factor target enrichment is an integral step in microarray and RNA-seq expression analysis. While numerous pipelines and tools exist to perform these types of analyses, they often require advanced knowledge of computational programming and statistical methods. Online tools and local packages exist to streamline expression and enrichment analysis, often at the expense of statistical rigor; while Over Representation Analysis (ORA) on differentially expressed genes (DEGs) can offer useful biological insight, tools which incorporate differential expression and geneset or transcription factor target enrichment for simplicity and accessibility for all users seldom offer the option for ranked-based enrichment approaches. Furthermore, reanalyzing existing transcriptomic datasets, particularly for RNA-seq data, can be computationally expensive. Therefore, in order to make reanalysis of existing datasets as well as advanced ranked-based and gene-regulatory analysis approaches more accessible, we developed DEGenR, an interactive Shiny application that provides computationally lightweight tools for identifying differentially expressed genes (DEGs) from user-provided RNA-seq read counts or publicly available microarray (NCBI GEO) and RNA-seq (recount2) datasets combined with pathway, ontology, and transcription factor enrichment analysis using multiple user-selected gene ranking approaches and geneset databases.
Materials and Methods
DEGenR is a Shiny application that runs from within a local instance of R. The source code and user guide for DEGenR are available at https://www.github.com/DEGenR. The current implementation depends on several R packages, including shiny, GEOQuery, recount, edgeR, limma, topconfects, msigdbr, tmod, camera, fgsea, enrichR, viper, and DoRothEA. Users can find a list of all packages and databases required by the DEGenR in the “packages_required.txt” file in the DEGenR folder download from GitHub. A sample guided workflow with a case study can be found in Supplementary File S1.
Results
For differential gene expression (DGE) analysis, DEGenR leverages the edgeR, limma, and topconfects R packages. The first step in the DEGenR analysis pipeline is the generation of the RNA gene or probe count matrix (Figure 1). Users can select or submit data in three ways: 1) selection of a Sequencing Read Archive (SRA) RNA-seq study processed by recount2 (Collado-Torres et al., 2017) by providing a SRP study accession number available at recount2 (https://jhubiostatistics.shinyapps.io/recount/), including GTEx and TCGA datasets; 2) selection of microarray study in the Gene Expression Omnibus (GEO) (Barrett et al., 2013), by providing a GEO study accession number; or 3) providing an RNA-seq raw count matrix and metadata information. Option 1 uses the R package recount to download the processed and summarized RNA-seq data (SummarizedExperiment) of the recount2 SRP study of interest. This dataset is then converted to a DGEList R object from the count and metadata information using the edgeR (Robinson et al., 2010) function SE2DGEList for downstream DGE analysis. For option 2, a user may select a GEO microarray study for analysis, whereby DEGenR will use the GEOQuery R package (Davis and Meltzer, 2007) to retrieve the expression matrix and metadata for DGE analysis. For option 3, user may provide two files as the starting point for analysis: a raw count file in the comma-separated value (.csv) format, with the first column providing the gene information (e.g. ENSEMBL transcript or gene ID, HUGO gene symbol) and all subsequent columns provide the read counts for each sample; and a tab-delimited metadata file defining the sample names and associated metadata (e.g. condition, treatment). A publicly available RNA-seq data, GSE76987, is included with the DEGenR installation as an example study for users to process and explore each analysis method. Once an RNA-seq dataset has been uploaded by the user, it is converted to a DGElist object using the function DGEList in edgeR and genes with low counts are filtered out using filterByExpr function in edgeR. Raw read counts are normalized using the calcNormFactors function, which calculates a set of factors that minimize the log-fold changes or composition biases between the samples. Following these pre-processing steps, DEGenR provides the user with multiple tables and plots for interpretation of sample quality: sample-level count summary of library sizes and normalization factors, density and box plot prior to and following the removal of low expression genes, and a multidimensional scaling (MDS) plot to illustrate the relationships between samples within conditions. Similarly, microarray datasets from GEO are log2 normalized (if required) and prepared for downstream analysis with similar sample expression plots.
Following sample pre-processing, DEGenR uses the R package limma for DGE analysis. If the user has selected an RNA-seq study, voom (Law et al., 2014) transformation is performed to prepare for linear modeling, and a mean-variance trend plot is generated. Next, users define the groups for comparison in order to create the contrast matrix. For GEO datasets, users have two ways to define the groups: entering the GEO accession IDs for each group; and entering the column name and subsequently the sample names for each group.
For testing the differential expression of genes, users can select between three methods: the eBayes and TREAT tests, both implemented in the limma R package, or the topConfects method from the R package of the same name. Once the user has selected a method for DGE analysis, DEGenR offers multiple methods for both ontology and transcription factor (TF) enrichment analysis. For ontology enrichment analysis, users can select between five different methods: the camera (Wu and Smyth, 2012) test implemented in the R package limma (Ritchie et al., 2015); the fast multilevel version of the original GSEA (Subramanian et al., 2005) algorithm implemented in the R package fgsea (Korotkevich et al., 2019; Subramanian et al., 2005); the CERNO and hypergeometric tests implemented in the R package tmod (Weiner 3rd and Domaszewska, 2016); and the Enrichr (Chen et al., 2013; Kuleshov et al., 2016) overrepresentation analysis (ORA) test using the R package enrichR. For the GSEA, hypergeometric, and Enrichr tests, the user can select any of the eBayes, TREAT (McCarthy and Smyth, 2009), or topConfects (Harrison et al., 2019) differential gene testing methods to derive ranking metrics for all genes (fGSEA test) or a subset of genes by selecting FDR adjusted p-value and log2Fold-Change (log2FC) or confect ranking cutoffs (hypergeometric and Enrichr tests). By default, the camera and CERNO tests use the eBayes t-value or the minimum significant difference (MSD) as the gene ranking metrics, respectively. For each test, users are also able to select a geneset database in the Molecular Signatures Database (Liberzon et al., 2011, 2015) (MSigDB v7.2) (camera, fGSEA, CERNO, and hypergeometric tests) or all current geneset databases in Enrichr (Enrichr test). For TF enrichment analysis, these five tests from ontology enrichment are similarly available with TF-gene target geneset database, including consensus ChIP-seq databases from Enrichr (ENCODE-ChEA Consensus) and CHEA3 (ReMap ChIP-seq and Literature ChIP-seq) (Chèneby et al., 2018; Keenan et al., 2019, 3), GTRD (Kolmykov et al., 2021), and TRRUST (Han et al., 2018), as well as miRNA-gene target databases (MSigDB miRNAs, TargetScan [Agarwal et al., 2015], and miRTarBase [Chou et al., 2018]). An additional method for TF activity analysis is provided, which uses the DoRothEA TF-gene regulon databases (Garcia-Alonso et al., 2019) (implemented in the DoRothEA R package) with the viper (Alvarez et al., 2016) algorithm to identify TFs with significantly changing activity between the two conditions. As with ontology enrichment analysis, users are able to select both the gene ranking metric and TF-gene gene set database for each method. A detailed workflow with case studies is available in Supplementary file S1.
In summary, with DEGenR shiny app, users start with count matrix and are able to perform different types of gene expression analysis, ontology enrichment, and transcription factor enrichment analysis along with table and plots provided for download at each corresponding step of the analysis.
Discussion
While there exists- thousands of human microarray and RNA-seq datasets, analyzing them can be both computationally expensive and requires considerate knowledge of R programming and analysis pipelines. Programs like GEOQuery and recount2 have made many of these microarray and RNA-seq datasets available pre-proccessed and parsed into an R format, while Shiny tools (Li and Andrade, 2017) and enrichment analysis websites have simplified DGE analysis and ontology and TF enrichment, respectively. However, this simplification often comes at the expense of statistically advanced approaches, particularly in the case of enrichment analysis; while ORA analysis can be powerful, ranked-based enrichment approaches frequently uncover biological insights missed by ontology and TF ORA enrichment analysis. DEGenR leverages the library of processed count-based transcriptomic datasets available via recount2 and GEOQuery or uploaded by a user and seamlessly connects them to multiple DGE test methods as well as multiple ranked-based enrichment tests for both ontology and TF genesets. As a result, DEGenR enables fast, advanced expression and enrichment analysis of thousands of sequencing studies to users of all experience levels.
Notes
Files
SubramaniamLab/DEGenR-v1.0.0.zip
Additional details
Related works
- Is supplement to
- https://github.com/SubramaniamLab/DEGenR/tree/v1.0.0 (URL)
References
- Agarwal,V. et al. (2015) Predicting effective microRNA target sites in mammalian mRNAs. eLife, 4, e05005.
- Alvarez,M.J. et al. (2016) Functional characterization of somatic mutations in cancer using network-based inference of protein activity. Nat. Genet., 48, 838–847.
- Barrett,T. et al. (2013) NCBI GEO: archive for functional genomics data sets—update. Nucleic Acids Research, 41, D991–D995.
- Chen,E.Y. et al. (2013) Enrichr: interactive and collaborative HTML5 gene list enrichment analysis tool. BMC Bioinformatics, 14, 128.
- Chèneby,J. et al. (2018) ReMap 2018: an updated atlas of regulatory regions from an integrative analysis of DNA-binding ChIP-seq experiments. Nucleic Acids Res., 46, D267–D275
- Chou,C.-H. et al. (2018) miRTarBase update 2018: a resource for experimentally validated microRNA-target interactions. Nucleic Acids Res, 46, D296–D302.
- Collado-Torres,L. et al. (2017) Reproducible RNA-seq analysis using recount2. Nature Biotechnology, 35, 319–321.
- Davis,S. and Meltzer,P.S. (2007) GEOquery: a bridge between the Gene Expression Omnibus (GEO) and BioConductor. Bioinformatics, 23, 1846–1847.
- Garcia-Alonso,L. et al. (2019) Benchmark and integration of resources for the estimation of human transcription factor activities. Genome Res., 29, 1363–1375.
- Han,H. et al. (2018) TRRUST v2: an expanded reference database of human and mouse transcriptional regulatory interactions. Nucleic Acids Res, 46, D380–D386.
- Harrison,P.F. et al. (2019) Topconfects: a package for confident effect sizes in differential expression analysis provides a more biologically useful ranked gene list. Genome Biol., 20, 67.
- Keenan,A.B. et al. (2019) ChEA3: transcription factor enrichment analysis by orthogonal omics integration. Nucleic Acids Res, 47, W212–W224.
- Kolmykov,S. et al. (2021) GTRD: an integrated view of transcription regulation. Nucleic Acids Research, 49, D104–D111.
- Korotkevich,G. et al. (2019) Fast gene set enrichment analysis. bioRxiv, 060012.
- Kuleshov,M.V. et al. (2016) Enrichr: a comprehensive gene set enrichment analysis web server 2016 update. Nucleic Acids Res., 44, W90-97.
- Law,C.W. et al. (2014) voom: precision weights unlock linear model analysis tools for RNA-seq read counts. Genome Biology, 15, R29.
- Li,Y. and Andrade,J. (2017) DEApp: an interactive web interface for differential expression analysis of next generation sequence data. Source Code for Biology and Medicine, 12, 2.
- Liberzon,A. et al. (2011) Molecular signatures database (MSigDB) 3.0. Bioinformatics, 27, 1739–1740.
- Liberzon,A. et al. (2015) The Molecular Signatures Database (MSigDB) hallmark gene set collection. Cell Syst, 1, 417–425.
- McCarthy,D.J. and Smyth,G.K. (2009) Testing significance relative to a fold-change threshold is a TREAT. Bioinformatics, 25, 765–771.
- Ritchie,M.E. et al. (2015) limma powers differential expression analyses for RNA-sequencing and microarray studies. Nucleic Acids Res, 43, e47–e47.
- Robinson,M.D. et al. (2010) edgeR: a Bioconductor package for differential expression analysis of digital gene expression data. Bioinformatics, 26, 139–140.
- Subramanian,A. et al. (2005) Gene set enrichment analysis: A knowledge-based approach for interpreting genome-wide expression profiles. PNAS, 102, 15545–15550.
- Weiner 3rd,J. and Domaszewska,T. (2016) tmod: an R package for general and multivariate enrichment analysis PeerJ Inc.
- Wu,D. and Smyth,G.K. (2012) Camera: a competitive gene set test accounting for inter-gene correlation. Nucleic Acids Research, 40, e133–e133.