search

weizhou0 / qtl

GNU General Public License v3.0
21 stars 2 forks source link

Questions of Cov #2

Closed Zepeng-Mu closed 5 months ago

Zepeng-Mu commented 7 months ago

Hi, thanks for developing this package! It is something that I've been waiting for so long. Since I have not used SAIGE before, I have several naïve questions on including Cov in the model.

  1. If I do not have related donors, i.e. not using GRM, should I include genotype PC as covariates to adjust for admixture (just as GWAS or bulk QTL)?
  2. If I should include genotype PC, should it be included as sampleCovarColList? Since all single cells from the same individual will have the same genotype PC, but I'm not sure if sampleCovarColList is modeled as random variable. If yes, may be I should not include continuous cov in this?
  3. For Step 1 in SAIGE-QTL, should I estimate variance ratio for each gene (feature) separately? It seems so in the tutorial. If yes, should I only include cis-SNP or genome-wide SNP in this step for each feature if I only want to map cis-QTL?

Thanks so much!!