estimateCellProp: Cell type proportion estimator

In xuz1/ENmix: Quality control and analysis tools for Illumina DNA methylation BeadChip

Cell type proportion estimator

Description

To estimates relative proportion of underlying cell types in a sample based on reference methylation data of pure cell types.

Usage

estimateCellProp(userdata,refdata="FlowSorted.Blood.450k",
                cellTypes=NULL,nonnegative = TRUE,nProbes=50,
                normalize=TRUE,refplot=FALSE)

Arguments

userdata	The input can be rgDataSet,methDataSet, MethylSet,RGChannelSet or methylation beta value matrix.
refdata	Reference data set will used. Current option: "FlowSorted.Blood.450k", "FlowSorted.DLPFC.450k","FlowSorted.CordBlood.450k", "FlowSorted.CordBloodCombined.450k", "FlowSorted.CordBloodNorway.450k" or "FlowSorted.Blood.EPIC".
cellTypes	Specific set of cell type data in reference data will be used for deconvolution. if "NULL" all cell types data will be used. see details for possible cell types
normalize	TRUE or FALSE, if TRUE, quantile normalization on methylated and unmethylated intensities will be performed.
nonnegative	TRUE or FALSE. If TRUE, the estimated proportions will be constrained to nonnegative values
nProbes	Number of best probes for each cell types will be used for the estimation.
refplot	TRUE or FALSE. IF TRUE, refdata distribution and heatmap will be plotted for inspection of reference dataset.

Details

This function use the method of Houseman et al (2012) to estimate cell type proportions based on reference DNA methylation data.

The following reference datasets can be used to assist the estimation. User should select a reference most resemble to user's data in tissue, age, and array type.

FlowSorted.Blood.450k: consisting of 450K methylation data for 60 blood samples from 6 male adults. Six samples for each of the cell types: Bcell CD4T CD8T Eos Gran Mono Neu NK PBMC WBC; See Reinius et al. 2012 for details.

FlowSorted.CordBlood.450k: consisting of 450k methylation data for 104 cord blood samples from 17 male and female individuals. Cell type (# samples) are: Bcell(15) CD4T(15) CD8T(14) Gran(12) Mono(15) NK(14) nRBC(4) WholeBlood(15). See Bakulski et al. Epigenetics 2016 for details.

FlowSorted.CordBloodNorway.450k: consisting of 450K methylation data for 77 cord blood samples from 11 individuals (6 girls and 5 boys). 11 samples for each of the cell types: Bcell CD4T CD8T Gran Mono NK WBC. See P Yousefi et al Environ. Mol. Mutagen 2015 for details.

FlowSorted.Blood.EPIC: consisting of EPIC methylation data for 37 magnetically sorted blood cell references from 12 individuals. See LA Salas et al. 2018 for details.

FlowSorted.DLPFC.450k: consisting of 450K methylation data for 58 brain tissue samples from 29 individuals. 15 females and 14 males, 6 Africans and 23 Caucasians, age range from 13 to 79. 29 samples for each of the cell types: NeuN_neg and NeuN_pos. See Guintivano et al. 2013 for details.

FlowSorted.CordBloodCombined.450k: consisting of 289 combined umbilical cord blood cells samples assayed by Bakulski et al, Gervin et al., de Goede et al., and Lin et al. see https://github.com/immunomethylomics/FlowSorted.CordBloodCombined.450k. details.

Value

The output is a data frame composed of the estimates of cell type proportions with columns indicate cell types and rows indicate samples.

Author(s)

Zongli Xu

References

EA Houseman, WP Accomando, DC Koestler, BC Christensen, CJ Marsit, HH Nelson, JK Wiencke and KT Kelsey. DNA methylation arrays as surrogate measures of cell mixture distribution. BMC bioinformatics (2012) 13:86.

Examples

require(minfiData)
path <- file.path(find.package("minfiData"),"extdata")
#based on rgDataset
rgSet <- readidat(path = path,recursive = TRUE)
celltype=estimateCellProp(userdata=rgSet,refdata="FlowSorted.Blood.450k",
         nonnegative = TRUE,normalize=TRUE)
#using methDataSet
qc=QCinfo(rgSet)
mdat<-preprocessENmix(rgSet, bgParaEst="oob", dyeCorr="RELIC",
                      QCinfo=qc,  nCores=6)
celltype=estimateCellProp(userdata=mdat,refdata="FlowSorted.Blood.450k",
        nonnegative = TRUE,normalize=TRUE)
mdat<-norm.quantile(mdat, method="quantile1")
#using beta value
beta<-rcp(mdat,qcscore=qc)
celltype=estimateCellProp(userdata=beta,refdata="FlowSorted.Blood.450k",
         nonnegative = TRUE)

xuz1/ENmix documentation built on Aug. 5, 2023, 7:11 a.m.