I am struggling what type of analysis would be appropriate in my situation. I am working on a rare disease (prevalence ~0.0004) and my sample size is 4376 (1094 of these are cases) I have SNParray data and I imputed SNPs using HRC as a reference and want to restrict to SNP with MAF>5% (because these will be most reliable imputed).
My understanding is that single component GREML are highly sensitive to assumptions (or is that only problematic when including SNP with MAF>5%), but multi-component GREML required large samples sizes.
The genetic architecture of my disease is unknown, but it is a severe, developmental disorder and thus I expect that a (large) proportion of heritability will also be rare variants..(something that will require another study design to test).
So should I go for GREML-SC to be able to answer the question how much of the variance can be attributed to common variants?
On top of that my sample exists of individuals from different countries (all European though, but there will be some population stratification, although I used matched controls). So would adding the pinciple components be necessary?