Learn R Programming

⚠️There's a newer version (1.0.22) of this package.Take me there.

ADAPTS

Augments existing or de-novo cell-type signature matrices to deconvolve bulk gene expression data This package expands on the techniques outlined in Newman et al.'s 2015 Nature Methods paper: 'Robust enumeration of cell subsets from tissue expression profiles'. to allow a user to easily add their own cell types (e.g. a tumor specific cell type) to Newman's LM22 or other signature matrix.

To install this package in R, use devtools.

install.packages('devtools')

library(devtools)

devtools::install_github('sdanzige/ADAPTS')

Data for the Vignette can be found at sdanzige/ADAPTSdata and sdanziger/ADAPTSdata2

More information about this package is available on bioRxiv (https://www.biorxiv.org/content/10.1101/633958v2) and this package has been officially released as an R package on CRAN (https://cran.r-project.org/web/packages/ADAPTS/).

Copy Link

Version

Install

install.packages('ADAPTS')

Monthly Downloads

508

Version

0.9.21

License

MIT + file LICENSE

Maintainer

Samuel Danziger

Last Published

August 7th, 2019

Functions in ADAPTS (0.9.21)

estCellPercent.nnls

Use non-negative least squares regression to deconvolve a sample This is going to be to simple to be useful This might be more interesting if I used non-positive least squares to detect 'other'

Leukocyte 22 data matrix

Load the MGSM27 signature matrix

estCellCounts.nPass

Deconvolve with an n-pass spillover matrix

Load the LM22 xCell map

estCellPercent.svmdecon

Use SVMDECON to estimate the cell count percentage Performs considerably worse in deconvolution than DCQ

estCellPercent.spillOver

Use a spillover matrix to deconvolve a samples

Use weightNorm to normalize the SVM weights. Used for SVMDECONV

hierarchicalClassify

Build clusters based on n-pass spillover matrix

Load a map of cell type names

Licenses required by Celgene legal

hierarchicalSplit

Attempt to deconvolve cell types by building a heriarchy of cell types using spillToConvergence to determine cell types that are not signficantly different. First deconvolve those clusters of cell types. Deconvolution matrices are then built to separate the cell types that formerly could not be separated.

Use the full LM22 data matrix and add a few additional genes to cover osteoblasts, osteoclasts, Plasma.memory, MM. In many ways this is just a convenient wrapper for AugmentSigMatrix

spillToConvergence

Build an n-pass spillover matrix, continuing until the results converge into clusters of cell types

Use a t-test to rank to features for each cell type

Use parallel missForest to impute missing values. This wrapper is required because missForest crashed if you have more cores than variables. This will default to no parellelization for Windows

estCellPercent.proportionsInAdmixture

Use R function proportionsInAdmixture to estimate the cell count percentage Uses the 'WGCNA' package

estCellPercent.DeconRNASeq

Use DeconRNASeq to estimate the cell count percentage Performs with similar effectiveness as DCQ, but identifies different proportions of cell-types Requires installation of package 'DeconRNASeq': source("https://bioconductor.org/biocLite.R") biocLite("DeconRNASeq")

Use SVMDECONV to estimate the cell count percentage David L Gibbs, dgibbs@systemsbiology.org June 9, 2017

estCellPercent.DCQ

Use DCQ to estimate the cell count percentage Requires installation of package 'ComICS' To Do: Also report the standard deviation as a confidence metric

buildSpilloverMat

Build a spillover matrix, i.e. what do purified samples deconvolve as?

Myeloma Genome Signature Matrix 27

clustWspillOver

Build clusters based on n-pass spillover matrix

AugmentSigMatrix

Build an augmented signature matrix from an initial signature matrix, source data, and a list of differentially expressed genes (gList) The user might want to modify gList to make certain that particular genes are included in the matrix The algorithm will be to add one additional gene from each new cell type Record the condition number, and plot those. Will only consider adding rows shared by fullData and newData

collapseCellTypes

Collapse the cell types (in rows) to super-classes #Updated 09-06-18 to include MGSM36 cell types