README.md
In sigven/oncoEnrichR: Functional annotation and enrichment analysis of genesets in the context of cancer

Overview
News
Annotation resources
Example report
Getting started
Contact

oncoEnrichR is an R package for functional interrogation of human genesets in the context of cancer.

The package is intended for exploratory analysis and prioritization of a gene list (referred to as query set below) from high-throughput cancer biology experiments, e.g. genetic screens (siRNA/CRISPR), protein proximity labeling, or transcriptomics (differential expression). The tool queries a number of high-quality data resources in order to assemble useful gene annotations and analyses in an interactive report. The contents of the final report attempts to shed light on the following questions:

Which diseases/tumor types are associated with genes in the query set, and to what extent?
Which proteins in the query sets are druggable in diffferent cancer conditions (early and late clinical development phases)? For other proteins in the query set, what is their likelihood of being druggable?
Which protein complexes are relevant for proteins in the query set?
Which subcellular compartments (nucleus, cytosol, plasma membrane etc) are dominant localizations for proteins in the query set?
Are specific tissues or cell types enriched in the query set, considering tissue/cell-type specific expression patterns of target genes?
Which protein-protein interactions are known within the query set? Are there interactions between members of the query set and other cancer-relevant proteins (e.g. proto-oncogenes, tumor-suppressors or predicted cancer drivers)? Which proteins constitute hubs in the protein-protein interaction network?
Are there specific pathways, biological processes or molecular functions that are enriched within the query set, as compared to a reference/background set?
Which members of the query set are frequently mutated in tumor sample cohorts (TCGA, SNVs/InDels, homozygous deletions, copy number amplifications)?
Which members of the query set are co-expressed (strong negative or positive correlations) with cancer-relevant genes (i.e. proto-oncogenes or tumor suppressors) in tumor sample cohorts (TCGA)?
Which members of the query set are associated with better/worse survival in different cancers, considering high or low gene expression levels in tumors?
Which members of the query set are associated with cellular loss-of-fitness in CRISPR/Cas9 whole-genome drop out screens of cancer cell lines (i.e. reduction of cell viability elicited by a gene inactivation)? Which targets are prioritized therapeutic targets, considering fitness effects and genomic biomarkers in combination?

March 31st 2021: 0.9.2 release
Added more protein identifier types (RefSeq peptide accession, Ensembl protein accession)
Updated KEGG pathway database (2021-03-29)
March 24th 2021: 0.9.1 release
Revised Disease Association section
- Genes in query set ranked according to overall cancer association strength
- Heatmap with tumor-type specific associations pr. gene
Updated datasets: MSigDB (v7.3)
March 21st 2021: 0.9.0 release
Added target tractability data (section Drug-target associations)
Added target priority scores (section CRISPR/Cas9 loss-of-function)
Added possibility to use Ensembl/RefSeq transcript identifiers as input
Added argument show_top_diseases_only - limits disease associations to top 15

Data harvested from the following resources form the backbone of oncoEnrichR:

Open Targets Platform - human drug-target associations and comprehensive disease-target associations
CancerMine - literature-mined database of drivers, oncogenes and tumor suppressors in cancer
The Cancer Genome Atlas - gene aberration frequencies and co-expression patterns in ~9,500 primary tumor samples
The Human Protein Atlas - expression data for healthy human tissues (GTex)/cell types, and prognostic gene expression associations in cancer
Molecular Signatures Database (MSigDB) - collection of annotated (e.g. towards pathways) genesets for enrichment/overrepresentation analysis. This includes genesets from Gene Ontology, Reactome, KEGG, WikiPathways, BIOCARTA, as well as curated immunologic and cancer-specific signatures.
NetPath - signaling transduction pathways
STRING - protein-protein interaction database
CORUM - protein complex database
ComPPI - subcellular compartment database
Project Score - Database on the effects on cancer cell line viability elicited by CRISPR-Cas9 mediated gene activation

Installation

install.packages('devtools')
devtools::install_github('sigven/oncoEnrichR')
library(oncoEnrichR)

Usage

oncoEnrichR performs its operations through the following procedures/methods:

1. oncoEnrichR::onco_enrich()

Consists of two main processing steps:

1) Takes an input/query list of human gene/protein identifiers (e.g. UniProt accession, RefSeq/Ensembl transcript identifer etc.) as input and conducts uniform identifier conversion

2) Performs extensive annotation, enrichment and membership analyses of the query set against underlying data sources on cancer-relevant properties of human genes and their interrelationships.
Technically, the method returns a list object with all contents of the analyses performed. The specific arguments/options and default values are outlined below:

r onco_enrich( query, query_id_type = "symbol", ignore_id_err = TRUE, project_title = "Project title", project_owner = "Project owner", project_description = "Project description", bgset = NULL, bgset_id_type = "symbol", bgset_description = "All protein-coding genes", p_value_cutoff_enrichment = 0.05, p_value_adjustment_method = "BH", q_value_cutoff_enrichment = 0.2, min_geneset_size = 10, max_geneset_size = 500, min_subcellcomp_confidence = 1, simplify_go = TRUE, ppi_add_nodes = 50, ppi_score_threshold = 900, show_ppi = TRUE, show_drugs_in_ppi = TRUE, show_disease = TRUE, show_top_diseases_only = TRUE, show_drug = TRUE, show_enrichment = TRUE, show_tcga_aberration = TRUE, show_tcga_coexpression = TRUE, show_subcell_comp = TRUE, show_crispr_lof = TRUE, show_cell_tissue = TRUE, show_prognostic_cancer_assoc = TRUE, show_complex = TRUE)

Argument |Description ------------- |---------------- query | query_id_type ignore_id_err project_title project_owner project_description bgset | bgset_id_type bgset_description p_value_cutoff_enrichment p_value_adjustment_method q_value_cutoff_enrichment min_geneset_size max_geneset_size min_subcellcomp_confidence| simplify_go ppi_add_nodes ppi_score_threshold show_ppi show_drugs_in_ppi show_disease show_top_diseases_only| show_drug show_enrichment show_tcga_aberration show_tcga_coexpression show_subcell_comp show_crispr_lof show_cell_tissue show_prognostic_cancer_assoc show_complex character vector with gene/query identifiers | character indicating type of identifier used for query set (one of "uniprot_acc", "symbol", "entrezgene", "ensembl_gene", "refseq_mrna", "refseq_protein", "ensembl_protein", or "ensembl_mrna") | logical indicating if analysis should continue when erroneous/unmatched query identifiers are encountered in query or background gene set | project title (report title) | project owner (e.g. lab/PI) | brief description of project, how target list was derived character vector with gene identifiers, used as reference/background for enrichment/over-representation analysis | character indicating type of identifier used for background set (one of "uniprot_acc", "symbol", "entrezgene", "ensembl_gene", "refseq_mrna", "refseq_protein", "ensembl_protein", or "ensembl_mrna") | character with description of background gene set (e.g. 'All lipid-binding proteins (n = 200)') | cutoff p-value for enrichment/over-representation analysis | one of "holm", "hochberg", "hommel", "bonferroni", "BH", "BY", "fdr", "none" | cutoff q-value for enrichment/over-representation analysis | minimal size of geneset annotated by term for testing in enrichment/over-representation analysis | maximal size of geneset annotated by term for testing in enrichment/over-representation analysis minimum confidence level of subcellular compartment annotations (range from 1 to 6, 6 is strongest) | remove highly similar GO terms in results from GO enrichment/over-representation analysis (recommended) | number of neighbouring nodes to add to query set when computing the protein-protein interaction network (STRING) | minimum significance score (0-1000) for protein-protein interactions to be included in network (STRING) | logical indicating if report should contain protein-protein interaction data of query set and their closely interacting partners (STRING), local network communities, and rank of query set based on network centrality | logical indicating if targeted drugs (>= phase 3) should be displayed in protein-protein interaction network (Open Targets Platform) | logical indicating if report should contain ranked associations to cancer phenotypes (overall), as well as tumor-type specific rankings (association score >= 0.4, minimum number of sources contributing to association >= 2, (Open Targets Platform)) logical indicating if report should only show top (20) cancer phenotypes/disease associations from Open Targets Platform | logical indicating if report should contain cancer drugs targeted towards proteins in the query list (early and late development phase) and tractability/druggability data for all query entries, from Open Targets Platform) | logical indicating if report should perform and list functional enrichment/over-representation analysis of query set (MSigDB, GO, KEGG, REACTOME, WikiPathways, NetPath) | logical indicating if report should contain TCGA aberration plots (amplifications/deletions, SNVs/InDels (oncoplots)) | logical indicating if report should list oncogenes/tumor suppressor genes that significantly correlate with entries in query set in terms of expression (across TCGA cohorts) | logical indicating if report should list subcellular compartment annotations (ComPPI) | logical indicating if report should list results from CRISPR/Cas9 loss-of-fitness screens and associated target priority scores (Project Score) | logical indicating if report should list results from tissue (GTex)- and cell-type (HPA) specific gene expression patterns in query set | logical indicating if report should list results from significant associations between gene expression and survival (Human Protein Atlas - Pathology Atlas) | logical indicating if report should show membership of target proteins in known protein complexes (CORUM)

2. oncoEnrichR::write()

Consists of two main processing steps:

1) Transform the contents of the analyses returned by oncoEnrichR::onco_enrich() into various visualizations and interactive tables

2) Assemble and write the final analysis report through
- A) a structured and interactive oncoEnrichR HTML report, OR
- B) a multisheet Excel workbook with results from the annotations and analyses provided by oncoEnrichR

Example run

A target list of n = 134 high-confidence interacting proteins with the c-MYC oncoprotein were previously identified through BioID protein proximity assay in standard cell culture and in tumor xenografts (Dingar et al., J Proteomics, 2015). We ran this target list through the oncoEnrichR analysis workflow using the following configurations for the onco_enrich method:

project_title = "cMYC_BioID_screen"
project_owner = "Raught et al."

and produced the following HTML report with results.

Below are R commands provided to reproduce the example output. NOTE: Replace "LOCAL_FOLDER" with a directory on your local computer:

library(oncoEnrichR)
myc_interact_targets <- read.csv(system.file("extdata","myc_data.csv", package = "oncoEnrichR"), stringsAsFactors = F)
myc_report <- oncoEnrichR::onco_enrich(query = myc_interact_targets$symbol, project_title = "cMYC_BioID_screen", project_owner = "Raught et al.")
oncoEnrichR::write(report = myc_report, file = "LOCAL_FOLDER/myc_report_oncoenrichr.html", format = "html")
oncoEnrichR::write(report = myc_report, file = "LOCAL_FOLDER/myc_report_oncoenrichr.xlsx", format = "excel")

sigven AT ifi.uio.no

OncoEnrichR is supported by the Centre for Cancer Cell Reprogramming at the University of Oslo/Oslo University Hospital, and Elixir Norway (Oslo node).