CellMixpackage.PDFVIP

An introduction to gene expression deconvolution and the CellMix package A Comprehensive Framework for Gene Expression Deconvolution Renaud Gaujoux June 4, 2013 Abstract This vignette motivates and describes the functionalities of the CellMix package, an R package for performing gene expression deconvolution analysis. The package de?nes a general framework to apply, develop and test gene expression deconvolution methods. It incorporates, generalises and extends the set of tools we implemented when developing a semi-supervised approach to this problem, and includes several other previously published algorithms, hence facilitating their application and comparison. A special focus is drawn on lists of cell/tissue marker genes, which are very valuable resources as they may be used not only as prior knowledge or post-hoc independent validation data by deconvolution algorithms, but also as gene sets in classical enrichment analysis. We envisage that this package will provide the bioinformatics research community with an easy to use and ?exible platform for working with deconvolution methods, and cell heterogeneity in omics data in general. This vignette aims at providing a general background on gene expression deconvolution, motivating the CellMix package, as well as describing its main features. It serves as supplementary material for the following article: Renaud Gaujoux et al. “CellMix: A Comprehensive Framework for Gene Expression Deconvolution”. In: accepted (2012) Documentation, practical examples, and sample analyses can be found online at: http://web.cbio.uct.ac.za/~renaud/CRAN/web/CellMix Note that to reproduce some of the examples in this vignettes, you will also need to have the GEOquery package1 package installed, as well as some Bioconductor annotation packages which will be installed when needed2: # install biocLite install.packages(BiocInstaller) # or alternatively do: source(?/biocLite.R?) # install GEOquery (NB: might ask you to update some of your packag