Heritability and genetic correlations

Heritability

Heritability measures the proportion of variance of the phenotype that is due to genetic differences between individuals. Heritability is always a property of a particular population, and its value can vary between different environmental conditions as the relative roles of genes and environment change.

Phenotypic correlations

Heritabilities have been estimated for a long time using phenotypic correlations between relatives of different degrees. For example, the traditional twin estimates compare phenotypic similarity in monozygotic twin pairs to similarity in dizygotic twin pairs. Under some strong assumptions (such as that the environmental contribution to the phenotypic similarity would be the same for a monozygotic as for a dizygotic twin pair, and that the genetic effects act additively over loci) it follows that the difference between the correlation estimates of these two types of twin pairs leads to an estimate of heritability

In practice, (narrow sense, that is, variance explained by the additive effects of the variants) heritability $h^2$ __ can be estimated by comparing individual phenotypic variation among related individuals in a population, by examining the association between individual phenotype and genotype data, or by modelling summary-level data from genome-wide association studies (GWAS).

Ldsc software

In FinnGen, ldsc software has been used to estimate heritabilities for all endpoints from the released summary statistics using both Finnish and European LD panel.

These heritability estimates can be found from:

/finngen/library-green/finngen_R8_analysis_data/ldsc/finngen_R(data freeze specific)_[FIN/EUR].ldsc.heritability.tsv,

and the documentation for the files can be found here.

Genetic correlation

A genetic correlation ($r_g$) between two traits is the proportion of variance that the two traits share due to genetic causes. A genetic correlation of 0 implies that the genetic effects on one trait are are independent of the other, while genetic correlation of 1 implies that all of the genetic influences on the two traits are identical. In practise, genetic correlations between two traits can be computed from the GWAS summary statistics by computing the correlation of the regression coefficients.

In FinnGen, ldsc software has been used to estimate genetic correlations for all endpoint pairs from the released summary statistics using both Finnish and European LD panel.

All these pairwise genetic correlations can be found from:

/finngen/library-green/finngen_R<ver>/finngen_R<ver>_analysis_data/ldsc/data/finngen_R<ver>_FIN.ldsc.summary.tsv

and the documentation for the files can be found here.

For more information:

please see this github page.