Post by mmyc3 on May 2, 2021 8:40:43 GMT
Many thanks for this great software.
I am trying to ascertain heritability using WGS for a rare trait (prevalence 0.0001) in which I have detected genome-wide significant loci using GWAS: I have 90 cases and 20,000 controls and have created 12 GRMs stratified by LD quartile and MAF (0.001-0.01, 0.01-0.05, >0.05). This cohort is European, although I have further samples that I wish to include from other ancestries. I used 10 PCs as covariates.
After running the REML analysis the observed heritability is sum V(G)/Vp is 0.31 with SE 0.05. The transformed liability threshold heritability sum V (G)_L/Vp is however 3.4 with SE 0.6.
My questions are:
1. Can I reliably use the software to calculate heritability in an admixed/mixed-ancestry population if I use the stratified MAF and LD GRM method as well as PCs as covariates?
2. How do I interpret the liability threshold heritability of 3.4? Why is this over 1 and not appear to be a proportion as seen with the observed heritability?
Many thanks in advance