What is "array heritability"?
Please point me to somewhere in the lectures/readings if I missed it.
Thanks!
Hi Jean,
It was at the very end of the lecture, on slide 29 -- I know we were a bit rushed towards the end, I'm sorry about that.
It's the heritability based on all the SNPs on the array, or all imputed variants. In the second lecture of the course you talked a little about estimating heritablity from family relationships, like sibling and parent-offspring relationships. This is a similar idea, using all variants to estimate (distant) genetic relationships amongst all individuals (kinship), and estimate heritability using these estimates. For example, parent-offspring share half their genome. Other individuals drawn from your study population share less of their genome, but are still distantly related.
As the slide highlights, it is restricted to narrow sense heritability, so it can differ from other heritability estimates for that reason (e.g., dominance, epistasis, as the example on slide 33 that we did not have time for), as well as it's limited to that which was genotyped/imputed. The family-based (i.e., siblings, parent-offspring) heritability estimates can also be somewhat inflated from shared environmental effects, which the array heritability estimates are often not as subject to (particularly when first and second degrees are excluded from them, as is often done with GCTA).
Software commonly used to calculate it include GCTA and LD Score if you are trying to find it for your trait of interest.
Best,
Tom
Thanks very much, and apologies for my oversight. This explanation is helpful.