cosDeco: Create consensus deconvolution.

In caanene1/Decosus: Method to generate robust consensus from published cell deconvolution/aggregation methods

Create consensus deconvolution.

Description

Function to apply seven deconvolution/signature methods.

Usage

cosDeco <- function(x=df, rnaseq=T, plot=TRUE, ext=FALSE,
                    sig=NULL, anno.1=NULL, anno.2=NULL,
                    cp=NULL, free=FALSE)

Arguments

x	The data frame with samples in column/names and rows/rownames as genes.
rnaseq	Boolen indicating if the x is the platform. Defaults to TRUE.
plot	Boolen to plot or not plot the evaluation correlation plots. Only need for development of the tool.
ext	Boolen indicating an extension to the in-built signatures, triggered by sig, anno.1, anno.2.
sig	Data frame of the extension signature. Needs at least anno.1 for conesus.
anno.1	Data frame of the annotation for signature.This is used for the relative conseus.
anno.2	Data frame of the annotation for signature.This is used for the absolute conseus.
cp	Data frame of the colour to use in the correlation plot. Relevant if plot is true, but not compulsory.
free	Boolen indicating if you want to remove methods with restrive licences. Currently affects EPIC.
scale.i	Boolen indicating if you want to individually scale outputs from each methods before combining.
agg.method	String of the aggregation method to use, either "mean" or "geomean".
mini.output	Boolen indicating if you want to extract just the final reuslts or everythin.

Details

A Consesus approache to cell proportions/deconvulations builds on the strength of aggregation and deconvolution methods. This generates robust estimates of cell proprotions/contents in a gene expression dataset. It provides a flexible paltform to add more signatures.

Value

A DecoCell class object, with input, intermediate results and a list "res.final" of two dataframes and two correlation pdfs in the working directory.

main_samples	Table of consesus values for comparing samples
main_cells	Table of consesus values for comparing cells
raw_results	Table of raw outputs of each method

Note

The packages associated with the methods are configured to install automatically. However, you can installl them directry. "devtools::install_github('dviraran/xCell', force = TRUE)" "devtools::install_github("ebecht/MCPcounter",ref="master", subdir="Source")" "devtools::install_github("GfellerLab/EPIC", build_vignettes = TRUE)". You need to install devtools.

Installation From [github](https://github.com/caanene1) with: devtools::install_github('caanene1/Decosus') From source in your working directory with: install.packages("Decosus_0.1.2.tar.gz", repos=NULL, type="source")

Please, read the github page for how to extened the signature and the required columns.

Author(s)

Chinedu A. Anene, PhD, Emma Taggart

References

Depends on: xcell MCP Danaher Davoli Rooney quanTISeq EPIC Uses base R functions.

See Also

The output of this function works well for the deconvolution of any tissue type. For cancer specific consesus, see the methods described by PMID: 31510660 or PMID: 31641033.

caanene1/Decosus documentation built on Feb. 24, 2024, 6:37 a.m.