Estimate cell composition in a public 450k sample using ewastools and the HRS coefficients

celltype_from_ewastools.R

# demo using the ewastools package to estimate leukocyte composition from a public 450k sample
# using results learned from HRS (see preprint https://www.biorxiv.org/content/10.1101/2022.11.28.518190v1.full)

if("ewastools" %in% rownames(installed.packages()) == FALSE){
  install.packages("BiocManager")
  BiocManager::install("illuminaio")
  remotes::install_github("hhhh5/ewastools") # at least v1.7.1
}

library(ewastools)
library(here)
library(R.utils)

# from https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=%20GSE87571
# grab idat files for first sample "X1 genomic DNA from whole blood"
filename <- "GSM2333901_9370847096_R05C02"
fileurlGrn <- "https://www.ncbi.nlm.nih.gov/geo/download/?acc=GSM2333901&format=file&file=GSM2333901%5F9370847096%5FR05C02%5FGrn%2Eidat%2Egz"
fileurlRed <- "https://www.ncbi.nlm.nih.gov/geo/download/?acc=GSM2333901&format=file&file=GSM2333901%5F9370847096%5FR05C02%5FRed%2Eidat%2Egz"

download.file(fileurlGrn, destfile = here(paste0(filename, "_Grn.idat.gz")))
download.file(fileurlRed, destfile = here(paste0(filename, "_Red.idat.gz")))
gunzip(paste0(filename, "_Grn.idat.gz"))
gunzip(paste0(filename, "_Red.idat.gz"))

# import and preprocess this sample
meth = 
	filename |>
	read_idats(quiet=FALSE) |>
	detectionP() |>
	correct_dye_bias() |>
	mask(0.05)
meth = dont_normalize(meth) # matrix of beta values

print(estimateLC(meth, ref = "HRS"), digits = 2)
# and for reference, here is a Houseman approach using Reinius as the reference dataset
print(estimateLC(meth, ref = "Reinius"), digits = 2)