Hi all,
I'm new to the words heritability, liability and familial risk in GWAS studies and I have three questions.
Firstly, I know the definition of heritability is the percentage of variation of phenotype captured by variation of genotype (correct me if I'm wrong). However, what's a liability here? I see some writing like "heritability on the liability threshold scale spanned a wide range" and do not understand.
Secondly, I'm reading some cancer GWAS papers and often at the end author would make a statement like "new discovered loci captured XX% of familial relative risk". Why don't they use heritability rather than the percentage of familial relative risk?
Finally, if the heritability of one phenotype is "a%" and on average "b%" of familial relative risk can be explained by a set of loci, does it mean I can do simple math that "a%-b%" of phenotype variation can be explained by the rest of genome, or something else?
Thanks,
Terry
For the first part, you can refer to Falconer 1996 Introduction to Quantitative Genetics.
Basically, the liability threshold model is used to explain binary outcome. For quantitative outcomes, you can get the variance of the phenotype, but you can't do the same for binary outcome. Falconer then proposed the liability threshold model, which state that diseases with binary outcome has an underlying continuous distribution, called the liability, which can be viewed as the additive effect of all genetics and non-genetics factors. When people have high enough liability (e.g. with liability passing through a threshold), they will develop a disease.
(I'm not familiar with familial relative risk)