laylagerami/consensustme

ConsensusTME Gene Sets and R Script

Consensus^TME: Tumour microenvironment cell estimation

The ConsensusTME package provides a integrative tool for R that uses a consensus approach to generating cancer specific signatures for multiple cell types found within the tumour microenvironment.

These consensus gene sets are then used within a ssGSEA framework to provide normalised enrichment scores for each of the cell types representing the relative abundance of cell types across multiple samples.

Installation

The recommended installation for the R package is to use the install_github function from the devtools package.

install.packages("devtools")
devtools::install_github("cansysbio/ConsensusTME")

Usage

Cell Type Estimation

library(ConsensusTME)

bulkExpMatrix <- as.matrix(read.delim(bulkGeneExpression.txt, row.names = 1))

ConsensusTME::consensusTMEAnalysis(bulkExpMatrix, cancer = "OV", statMethod = "ssgsea")

Example shows running Consensus^TME for ovarian samples where bulkExpMatrix is a numerical matrix with HUGO gene symbols as row names.

Currently Consensus^TME can be run for the following TCGA cancer types viewable through ConsensusTME::cancerAll:

ACC BLCA BRCA CESC CHOL COAD DLBC ESCA GBM HNSC KICH KIRC KIRP LGG LIHC LUAD
LUSC MESO OV PAAD PCPG PRAD READ SARC SKCM STAD TGCT THCA THYM UCEC UCS UVM

The statMethod argument gives the option to run Consensus^TME gene sets with the following statistical frameworks:

ssgsea singScore gsva plage plage zscore

Further information can be found in the documentation ?consensusTMEAnalysis

Run Gene Set Enrichment Separately

To use the same statistical framework as Consensus^TME with a chosen signature from the consensus compendium.

bindeaGeneSet <- ConsensusTME::methodSignatures$Bindea

ConsensusTME::geneSetEnrichment(bulkExpMatrix, bindeaGeneSet)

Generate Consensus^TME Gene Sets

Consensus^TME gene sets can be generated for other uses. N.B. These gene sets are curated to be good signatures for immune cells within the tumour microenvironment. For immune cells in a different biological context other signatures may be more appropriate.

Pre-processed Consensus^TME gene sets can be accessed with ConsensusTME::consensusGeneSets

rawMethodSignatures <- ConsensusTME::methodSignatures

matchedSigs <- ConsensusTME::matchGeneSigs(rawMethodSignatures)

ConsensusTME::buildConsensusGenes(matchedSigs)

Notes

• It is important to note that results generated in this way will provide quantification of cell types that are relative across samples rather than across cell types.

• Consensus^TME gene sets have been made to be cancer specific, if analysis is aimed at healthy tissue or PBMCs better results may be obtained by using alternative gene sets or methodologies.

• Consensus^TME is by design an evolvable tool. As more gene sets and signature matrices are made available we will assess whether including will improve performance and release a new version subsequently.

• Please contact any of the authors if you believe there are signatures or approaches that Consensus^TME would benefit from including.

• A comprehensive and unbiased set of benchmarking experiments were performed to test this approach. A shiny app for exploring results and further information can be found at: http://consensusTME.org

Publication

A manuscript is currently under consideration for publication, to cite currently please refer to the bioRxiv preprint:

https://www.biorxiv.org/content/early/2018/10/11/437533