Single quantitative variable as an outcome (not gene expression based)

[Rubbert]

I would like to run BootstrapQTL with as an outcome variable a single quantitative protein abundance "vector" (in this case the concentration of a particular cytokine).

For my "normal" calculations I use MatrixTQL as follows:

matrix_cQTLS = Matrix_eQTL_engine( snps = snpsSliced, gene = cytoSliced, cvrt = cvrtSliced, output_file_name = output_file_name, pvOutputThreshold = pvOutputThreshold, useModel = useModel, errorCovariance = numeric(), verbose = FALSE, pvalue.hist = FALSE, min.pv.by.genesnp = FALSE, noFDRsaveMemory = FALSE)

Where,

• snpsSliced contains the genetic SNP data,
• cytoSliced contains the protein concentation:
  Number of columns: 302 Number of rows: 1
• cvrtSliced = covariates
• pvOutputThreshold = 1
• useModel = LINEAR

Next, I would like to correct for the overestimation of effect sizes, and BootstrapQTL seems like a potential solution.
However, I am having trouble getting the settings right for this type of analysis, since it is designed for eQTL analyses.

I am currently doing the following:

  snpspos = as.data.frame(matrix(NA,ncol=3,nrow = nrow(snpsSliced)))
  colnames(snpspos) = c('snpid','chr','pos')
  snpspos$snpid = rownames(snpsSliced)
  snpspos$chr=1
  snpspos$pos=10000
  
  colnames(genepos) = c('geneid','chr','left','right')
  genepos$geneid = rownames(cytoSliced)
  genepos$chr=1
  genepos$left=snpspos$pos[[1]]-5000
  genepos$right=snpspos$pos[[1]]+5000
  
  test1 = BootstrapQTL(snps = snpsSliced, 
                       gene = cytoSliced, 
                       snpspos, 
                       genepos, 
                       cvrt = cvrtSliced,
                       n_bootstraps = 100,
                       local_correction = "none",
                       correction_type="weighted",
                       useModel=useModel,
                       errorCovariance = numeric())

Where I am basically artificially defining a random position for my "gene" (i.e. protein), and having all SNPs fall inside this, so they are "cis". However, the "get_eSNP_threshold" then gives an NA, since I only has one "gene".

I was wondering what settings you would recommend that would make sense in this context and that runs successfully, to correct for the overestimation of effect sizes in a valid way?

I hope this makes sense, and falls within the use-cases of what BootstrapQTL can correct for.

Thanks!

Rob