CRAN Task View: Genomics, Proteomics, Metabolomics, Transcriptomics, and Other Omics

In this task view, we focus on the most-important CRAN packages dedicated to omics data analysis including those related to annotation and databases. The term omics concerns various biological disciplines ending with the suffix -omics, such as genomics, proteomics, metabolomics, transcriptomics. We include CRAN packages which have been published more than one year ago and are regularly updated. Therefore, very recent packages may not appear yet but may be included soon if they are maintained. A few additional packages from the Bioconductor project are included, not looking for an exhaustive description but pointing to the must-see packages or to packages focused on omics not well covered by CRAN packages. Complementary information might also be found in Phylogenetics, MachineLearning, Agriculture or MissingData task views. Note that packages covering statistical genetics data (DNA data, haplotype estimation, population structure, genetic epidemiology), phylogenetics and phylogenomics are not covered by this task view. Still further packages might be found on Bioconductor, R-Forge, and GitHub.

If you think we have missed some important packages in this list, please e-mail the maintainers or submit an issue or pull request in the GitHub repository linked above.

The task view is structured into main topics:

• Annotation and databases
• Genomics
• Transcriptomics
• Proteomics
• Metabolomics
• Other omics
• Multiple omics
• Specific tasks
• Specific application fields

Annotation and databases

• Databases: geneExpressionFromGEO can be used to easily download a gene expression dataset from GEO. cbioportalR makes available clinical and genomic data from cBioPortal and TCGAretriever helps specifically to access and download TCGA data. pinfsc50 provides genomic data for the plant pathogen Phytophthora infestans.
• Functional annotation: enrichR (archived) provides an interface to the Enrichr databases. biomartr provides an interface to the BioMart database. msigdbr provides an interface for the MSigDB for Gene Set Enrichment Analysis. gprofiler2 provides an interface for g:Profiler (including functional enrichment analysis, identifier conversion, or orthology across species).
  toprdata provides gene and exon information from Ensembl genome builds. read.gb can read records with .gb extension from the NCBI Nucleotide database. Map2NCBI provides information on markers described by their positions by querying the NCBI database. PSSMCOOL contains the computation of various features from Position Specific Scoring Matrix (PSSM).
• Gene symbol conversions: aliases2entrez converts human gene symbols to curated gene entrezID from NCBI database. AnnotationBustR extracts subsequences into FASTA files
  from GenBank annotations where gene names may vary among accessions. BED is an interface for the Neo4j database providing a mapping between different identifiers of biological entities. More specifically, riceidconverter can be used to convert rice biological identifiers.
  babelgene converts between human and non-human gene orthologs/homologs from multiple databases.
  hgnc downloads and imports the HUGO gene nomenclature committee on mapping of gene symbols to gene entries in other databases.
• Gene set/pathway enrichment: WebGestaltR uses WebGestalt to perform Gene Set Enrichment and Network Topology analysis. GOxploreR provides a method to explore the gene ontology. GSA also propose methods for gene set enrichment analysis. VAM proposes a gene set testing method that is better designed than standard ones to handle single-cell RNA-seq data. tmod provides functions for gene set enrichment analysis in transcriptomic and metabolic profiling data. sigora corrects over-representation biases in pathway enrichment analysis. permPATH performs pathway enrichment analysis with permutation tests. CePa aims to find significant pathways through network topology information and pathfindR (accompanied by the data package pathfindR.data) and netgsa also use network information for enrichment analysis. SlaPMEG performs pathway enrichment analysis for longitudinal omics data. topologyGSA performs gene expression data tests using given pathway information on genes. diffEnrich compares functional enrichment between two experimentally-derived groups of genes or proteins. DysPIA identifies dysregulated pathways based on a pre-ranked gene pair list, using the DysPIAData.

Genomics

• Sequence manipulation and analysis: GenomicTools.fileHandler is a collection of I/O tools for handling the most commonly used genomic datafile types. valr can be used to read and manipulate genome intervals and signals. KRIS provides various routine functions for bioinformatics analyses of sequences. bioseq is a toolbox for manipulating biological (DNA, RNA and amino acid) sequences. ftrCOOL extracts features from nucleotide and peptide sequences and converts them to discrete numbers in order to be used as predictors in machine learning models.
  agvgd predicts “missense” substitutions based on the properties of amino acid side chains and protein multiple sequence alignments. vhica can be used to detect horizontal transfers of transposable elements from their divergence compared with regular genes.
  AnaCoDa contains a collection of models to analyze genome scale codon data using a Bayesian framework.
  BASiNET implements a method to classify RNA Sequences using network theory. MINTplates is a framework dedicated to exploration and annotation of tRNA and other genomic sequences. cumSeg estimates the number and location of change points in mean-shift (piecewise constant) models, such as genomic sequences.
• Cell analyses: sonicLength estimates the abundance of cell clones from DNA fragments. Similarly, scoper performs spectral clustering to identify clones from B cell data. Signac is a framework for the analysis and exploration of single-cell chromatin data. popPCR allows the classification of droplets in fluorescence activated cell sorting data to estimate DNA target concentration.
• Copy number data: ioncopy implements calculation of copy-numbers in amplicon sequencing data and ACNE provides allele-specific copy number from microarrays using NMF. aroma.cn implements several methods for normalizing and analyzing DNA copy-number data and PSCBS focuses on the analysis of parent-specific DNA copy-number data.
• Mutation and CRISPR: sigminer computes alteration signatures from genomic alteration records, Rediscover identifies mutually exclusive mutations using a Poisson-Binomial model, and mutSignatures identifies mutation signatures from somatic mutational catalogs. More generally, ICAMS analyses mutational signatures. Also, based on mutation data, ICBioMark identifies immunotherapy biomarkers. pathwayTMB estimates tumor mutational burden (TMB) with a new pathway-based gene panel. CB2 provides functions for CRISPR pooled screen analysis using Beta-Binomial Test.

Transcriptomics

Microarray or other continuous expression data

• Quality control and normalization: POD computes the probability of detection (POD) curve and the limit of detection (LOD). NACHO and nanostringr provide tools for quality control or normalization of NanoString nCounter data. MiRNAQCD is a toolbox for QC control of miRNA expression data.
• Integrated tools (with GUI): maGUI provides a graphical user interface to analyze microarray data (including annotation, tests, or network inference). depthTools (and its R commander plugin RcmdrPlugin.depthTools) is a collection of statistical tools based on data depth for gene expression analysis. OOMPA provides a collection of CRAN packages for microarray and proteomics analysis. seeker provides a standard interface and parallelization of various tools for transcriptomics such as BrainArray, RMA, GEOquery, ArrayExpress.
• Visualization: tinyarray is dedicated to the visualization of GEO and TCGA expression data. mpm provides exploratory graphical analysis of gene expression data with various factorial approaches (including PCA).
• Differential analysis: limma is among the most-important packages for microarray differential analysis. In addition, TailRank provides a tail-rank non parametric test for microarray datasets. SMVar implements structural model for variances for differential analysis of gene expression data. RNentropy implements a method based on information theory to detect significant variation in gene expression. optBiomarker estimates the optimal number of biomarkers in 2-group assays. leapp proposes a latent effect adjustment method to counter their effects on the rankings of hypotheses in tests. MCMC.qpcr implements Bayesian approaches for normalization and differential analysis of qPCR data. From a more global point of view, directPA can identify combinatorial effects of multiple treatments/conditions on pathways from different omics, including microarray and RNA-seq data.
  ssize.fdr provides functions to calculate appropriate sample sizes for gene expression tests and NewmanOmics implements Newman studentized range statistics for genome scale transcriptomics.
• Time series: TcGSA and TGS implement methods for longitudinal gene-expression data analysis. GeneCycle also analyses gene expression time series to detect periodically expressed genes. BClustLonG implements a Dirichlet process mixture model for clustering longitudinal gene expression data.
• Clustering: slfm performs gene expression analysis with a Bayesian sparse latent factor model. lmQCM implements a graph-based method for gene co-expression module discovery.
• Prediction: LPS implements the linear prediction score approach for gene expression signatures. SMDIC performs the identification of somatic mutation-driven immune cells from expression data. cubfits uses sequence and expression data to predict protein production rates.

RNA-seq data

• Generic: DESeq2 and edgeR are among the most-important packages for RNA-seq preprocessing and differential analysis.
• Mapping: MAAPER assigns 3’ RNA-seq reads to polyA sites.
• Deconvolution: BisqueRNA and InteRD provide methods to estimate cell type abundances from bulk expression data. scBio also proposes a deconvolution algorithm for bulk RNA-seq data but requires one single-cell reference sample. More specifically, imsig estimates the abundance of immune subpopulation cells in solid tumours. Finally, omicwas tests association with phenotypes that are cell-type specific in bulk omics experiments.
  
• Generic tool: Tmisc is a collection of utility functions to manipulate gene expression data.
• Simulation: seqgendiff provides a framework to simulate RNA-seq data under various assumptions and SeqNet and graphsim both offer simulations of RNA-seq data based on regulatory networks.
• Differential analysis: PQLseq implements a mixed model to account for population structure in RNA-seq (and other count data) differential analyses and glmmSeq also uses a mixed model for repeated measurements in transcriptomic datasets. NBPSequses Negative Binomial Beta tests for differential analyses of RNA-seq data. DiPALM enables differential analysis of time-course gene expression in different conditions. RVA is dedicated to the visualization of RNA-seq data and especially of the results of differential analysis.
  Similarly, markerpen uses penalized PCA for biomarker discovery. After differential analysis, DGEobj and DGEobj.utils provide a flexible container and a function toolkit to manage and annotate Differential Gene Expression (DGE) analysis results.
• Clustering: HTSCluster contains a model based on Poisson mixture to cluster RNA-seq datasets.
• Misc: SIBERG implements a method to identify bimodally expressed genes. CeRNASeek provides several functions to identify and analyse miRNA sponge.

Single-cell RNA-seq

• Generic tools: Seurat, iCellR and pagoda2 contain a collection of functions for single-cell data analysis, including differential analysis (the first one is based on the structure of SeuratObject). sccore and singleCellHaystack also provide methods for single-cell differential analysis, the second being based on KL divergence.
• Quality control: SoupX provides a method to remove mRNA contamination from single-cell RNA-seq data.
• Cell clustering and annotation: conos and scINSIGHT can be used to identify recurrent cell clusters in collections of single-cell RNA-seq datasets obtained in various conditions. ADAPTS constructs cell-type signature matrices using flow sorted or single cell samples and deconvolve bulk gene expression data. FiRE can discover rare cells from voluminous single cell expression data.
  scSorter assigns cells to known cell types according to marker genes and SignacX uses neural networks to identify cell types. clustermole can be used to identify human and mouse single-cell transcriptomic data cell type. immunarch can be used to analyze specifically T-cells and
• B-cells.
• Biomarker discovery: DIscBIO is a user-friendly multi-algorithmic pipeline for biomarker discovery in single-cell transcriptomics.
• Cell trajectories: dynwrap, treefit, and SCORPIUS infer cell trajectories from single-cell gene expression data. phateR can be used to visualize single-cell data with trajectories.
• Misc: rPanglaoDB can download and merge single-cell RNA-seq data from the PanglaoDB.

Proteomics

• Generic tool: ChemoSpec contains a collection of functions for top-down exploratory chemometrics for various proteomics spectroscopy data.
• Relation between proteins and DNA: geno2proteo allows to find the DNA and protein sequences of any given genomic loci using the ENSEMBL annotations. PredCRG contains a computational model to predict proteins encoded by circadian genes.
• Protein structure and sequence: compas manipulates and analyzes 3-D structural geometry of Protein Data Bank (PDB) files. bio3d contains tools to process, organize and explore protein structure, sequence and dynamics data. ampir is a toolkit to predict antimicrobial peptides from protein sequences on a genome-wide scale. ptm contains functions for the analysis of post-translational modifications in proteins. pbm contains various models to analyze protein-ligand interactions.
• Mass spectrometry proteomics: wrProteo contains a collection of functions for the analysis of mass spectrometry proteomic data. aLFQ implements the most common absolute label-free protein abundance estimation methods for LC-MS/MS and iq also implements protein quantification for mass spectrometry. protViz contains functions to visualize and analyze small mass spectrometry proteomic datasets.
  PTXQC and ypssc are both designed to analyze outputs of MaxQuant (a quantitative proteomics tool for large mass-spectrometric datasets). protti can also analyze outputs of MaxQuant in addition to outputs of Spectronaut or of Proteome Discoverer.  prozor determines the minimal protein set explaining peptide spectrum matches. net4pg implements a method to handle ambiguity for the identification in shotgun proteomics.
• Imputation: Several packages implement functions for the imputation of proteomics datasets: imputeLCMD (left-censored imputation), imp4p (including estimation of the missing value mechanism), and mi4p (multiple imputation). See also the MissingData view.
• Visualization: cp4p provides calibration plots for proteomics to check assumptions of FDR control procedures and to compute adjusted p-values. LSPFP provides plots of peptides from shotgun proteomics analysis.
• Prediction: krm is a package dedicated to kernel based regression methods that includes a kernel for protein sequence.
• Single-cell protein data (CITE-seq): dsb contains functions to normalize and denoise droplet single-cell proteomics data.

Metabolomics

• Data: metaboData contains examples of various metabolomics datasets.
• Generic tools: omu provides metabolomics analysis tools (including visualization) and lilikoi is a set of tools for metabolomics, including normalization and various prediction methods. MetabolomicsBasics also includes basic functions (peak picking, deconvolution, …) to explore metabolomics data.
• Identification: erah can be used to identify metabolites by spectral library matching in GC-MS untargeted metabolomics.
• Tests and biomarker discovery: iCARH implements a method for the analysis of time course metabolomic datasets, including biomarker discovery. MetabolicSurv contains biomarker discovery and survival analysis methods for metabolomics signatures. MetSizeR computes sample size for target statistical power in metabolomics experiments.
• Clustering: RAMClustR includes a clustering algorithm for mass spectrometry metabolomic data.
• Flux data: fbar is a toolkit for flux balance analysis and related metabolic modeling techniques.

Other omics

• Methylation: BiasCorrector is a GUI to correct measurement bias in DNA methylation analyses. rBiasCorrection performs bias correction in methylation data. TCA can deconvolve bulk condition-specific DNA methylation data into condition and individual specific methylation levels and detect associations with phenotypes.
• Hi-C: InteractionSet is a Bioconductor package providing a data structure adapted to Hi-C data. HiTC is a package to manipulate and analyze Hi-C data and diffHic is focused on differential analysis of Hi-C experiments.
• Lipidomics: interep can perform interaction analysis in high dimensional lipidomics datasets with repeated measurements. flippant allows the analysis of the activity of the lipid scrambling activity based on a fluorescence assay (dithionite scramblase assay).
• Bacterial genomics, microbiome and metagenomics: phyloseq is among the most-important packages for the analysis of high-throughput microbiome data. BarcodingR performs species identification using DNA barcodes. enveomics.R contains a collection of functions for microbial ecology and other applications of genomics and metagenomics and is the companion package for the Enveomics collection.
• AIRR-seq / Rep-seq: alakazam provides methods for high-throughput adaptive immune receptor repertoire sequencing (AIRR-Seq; Rep-Seq) analysis.

Multiple omics

• Generic: MOFA2 and mixOmics are among the most two important packages for integration multi-omics data sets.
• Single omics analysis: wrMisc contains a collection of tools to manipulate omics data and to perform various statistical analyses (including normalization and some statistical tests). integIRTy provides a method to identify genes that are consistently altered in cancer across different omics.
• Pathway analyses: ActivePathways combines p-values to obtain enriched pathways and processes. ICDS identifies pathways dysfunctional in cancer based on the integration of multiple omics.
• Exploratory integrative analyses: o2plsda provides functions to perform O2PLS-DA for multi-omics data integration. Similarly, CovCombR can be used to combine heterogeneous data sets through a covariance based method, and PMA proposes a sparse canonical correlation analysis to integrate multiple omics datasets. CovCombR and packMBPLSDA contain other methods to combine heterogeneous data sets through a covariance based method (the second being based on PLS-DA), and MOSS (archived) provides a sparse SVD-based multi-omics integration method. semmcmc also provides an omics integration method based on structural equation modelling (SEM), and IMIX uses Gaussian mixtures for multi-omics data integration. solvebio is a binding for the SolveBio API, a biomedical knowledge hub oriented toward omics data integration.
  More specifically, desiR provides functions for ranking, selecting, and integrating genes, proteins and metabolite data.
• Prediction: prioritylasso performs prediction from multiple omics using successive Lasso models using different priorities for the omics.
• Clustering: PINSPlus implements a robust clustering method for omics data integration, LUCIDus includes a ML based method for integrated clustering, and RHclust propose a k-means based multi-omics clustering.
• Meta-analyses: metaRNASeq and metaMA implement p-value combination techniques for meta-analysis of RNA-seq data and microarray data respectively. MetaIntegrator (archived) provides a pipeline for the meta-analysis of gene expression data.

Specific tasks

Quality control and normalization

• lineup and lineup2 include tools for detecting and correcting sample mix-ups in omics data.

Peak calling and analysis in data like ChIP-seq and ATAC-seq

• PeakSegOptimal implements a change point detection method based on the Poisson distribution for count data that includes a constraint suited for peak calling, PeakSegDisk provides a large scale implementation of the method using on-disk storage. PeakSegJoint implements a joint peak detection and PeakError computes true and false positive in peak calling with respect to annotated region labels.

Multiple testing

• bayefdr implements Bayesian estimation and optimization of expected False Discovery Rate and hommel includes methods for close testing with Simes’ inequality. rSEA performs simultaneous enrichment analysis that controls the FWER. More specifically, RobustRankAggreg provides a method for aggregating rank lists, especially lists of genes, DiscreteFDR implements multiple testing procedures adapted for discrete tests.

High-dimensional data regularization

• whitening implements whitening methods and CCA for high-dimensional omics data. mpmi uses a kernel smoothing approach for comparison of pairs of variables in large genomic datasets. hsstan uses a hierarchical Bayesian approach for biomarker discovery in high-dimensional datasets. supclust implements methods for supervised clustering of potentially many predictor variables (including ‘PELORA’ and ‘WILMA’).

Networks

• Gene network inference: WGCNA implements gene network inference with correlation based methods and GeneNet implements gene network inference with Gaussian Graphical Models. RGBM implements bootstrap based algorithms for network inference from microarray and RNA-seq data. RNAseqNet implements network inference with a log-linear Poisson model and can handle missing individuals. Patterns contains tools to infer biological networks with approaches designed for single or multiple joint omics. parmigene performs network inference with mutual information methods. networkABC performs network inference with Approximate Bayesian Computation. JSparO implements joint sparse optimization for gene network inference for cell fate conversion. Cascade implements a modeling tool allowing gene selection, reverse engineering, and prediction in cascade networks with experimental data provided in CascadeData. miic can be used to infer causal and non causal networks using information theory.
• Single-cell data: scLink and scTenifoldNet can be used to infer co-expression networks from single-cell data, the latter including an approach to compare these networks between different conditions.
• Differential analysis: scTenifoldKnk uses co-expression networks to identify differentially regulated genes by a virtual knockout approach and SIMMS enables integration of molecular profiles with functional networks (such as PPI networks) to detect biomarkers from survival data. Similarly, regnet provides network-based regularized models to perform variable selection in high-dimensional biological data. More generally, DiffCorr implements a method for identifying pattern changes between 2 experimental conditions in correlation networks. GANPA is a network-based gene weighting algorithm for pathway enrichment analysis and has a GANPAdata companion datasets package. DRaWR is a network-based method for ranking genes or properties related to a given gene set.
• PPI networks: prioGene can be used to define disease specific PPI networks and to deduce candidate gene prioritization.
• multi-omics networks: SmCCNet builds phenotype-specific multi-omics using a canonical correlation based framework.

Clustering

• EMMIXgene implements a mixture model-based approach for the clustering of microarray expression data and ORIClust performs clustering of short time-course or dose-response microarray gene expressions. OptCirClust performs clustering for circular data (like circular DNA or RNA molecules) and adjclust performs constrained clustering adapted to genomic constraints. The latter includes a wrapper for Hi-C datasets.

Visualization

• Sequence: JBrowseR provides an R interface to the JBrowse 2 genome browser. chromoMap provides interactive genomic visualization of the chromosomes or chromosome regions of any living organism. Similarly, seqinr provides visualizations for biological sequence (DNA and protein) data.
• Genome scale data: genoPlotR produces various gene or genome map figures ready for publications. Similarly, valr can be used to visualize genome-scale data, RCircos includes a collection of 2D circos plots for genomic visualization, and RIdeogram provides functions to display genome-wide data on ideograms. More specifically, gggenes draws gene arrow maps with a ggplot2 approach. VALERIE enables the visualization of alternative splicing events from single-cell data.
• Results of biomarker discovery methods: func2vis provides visualization and clean-up of enriched gene ontologies (GO) terms, protein complexes and pathways using ConsensusPathDB. statVisual implements visualizations for translational medicine and biomarker discovery. Similarly, volcano3D provides 3D volcano and polar plots that is suited to visualize biomarker differential analysis results for 3-class problems.
• Multiple omics: OmicNavigator is dedicated to the visualization of omics data. wilson is a web-based tool dedicated to the visualization of multi-omics data in an interactive way.
• Misc: BioInsight filters and plots the abundance of different RNA biotypes present in a count matrix.

Missing values

• SurrogateRegression performs estimation and inference on a partially missing target outcome while borrowing information from a correlated surrogate outcome.

Specific application fields

• Cancer: oncoPredict provides biomarker discovery from cell line screening data. tidyestimate infers tumor purity from expression data and Allspice provides a trained classifier for profiling patients with acute lymphoblastic leukemia (ALL) from RNA. driveR is a tool for personalized or batch analysis of genomic data for cancer driver gene prioritization by combining genomic information and prior biological knowledge. DRomics is dedicated to omics data obtained using a dose-response design, with a large number of tested doses.
  
• Nutrition: BRINDA computes biomarkers reflecting inflammation and nutritional determinants of anemia.

CRAN packages

Related links

• The Bioconductor project

Other resources

• CRAN Task View: Agriculture
• CRAN Task View: MachineLearning
• CRAN Task View: MissingData
• CRAN Task View: Phylogenetics
• Bioconductor Package: DESeq2
• Bioconductor Package: diffHic
• Bioconductor Package: edgeR
• Bioconductor Package: HiTC
• Bioconductor Package: InteractionSet
• Bioconductor Package: limma
• Bioconductor Package: mixOmics
• Bioconductor Package: MOFA2
• Bioconductor Package: phyloseq