Table 2. Singleton contributions to heritability.

Singletons vs. common variants (MAF>0.4)	Estimates	Models
	$\widehat{{\beta }^2}$ from ASE	${\beta }^2\propto {\left[pq\right]}^{-1}$	${\beta }^2$ independent of p
Proportion ${\sigma }_{g;\text{ASE}}^{2\widehat{}}$ explained by singletons	4.7%	36%	2.4%
Relative squared effect size	∼1.8	∼60	1
Selection strength	Weak	Strong	None

Estimates based on GTExV6 ASE data (cis-regulatory region +− 10kb from TSS) “$\widehat{{\beta }^2}$ from ASE” shows estimates of the median proportion of ${\sigma }_{g;\text{ASE}}^{2\widehat{}}$ explained by singletons based on effect sizes estimated from our ASE-model. “${\beta }^2\propto {\left[pq\right]}^{-1}$” shows the median expected per-gene heritability explained by singletons if singletons and common variants contributed equally to genetic variance. This is driven by the fact that 36% of all variants (median 36% of variants per gene) are singletons. The large relative effect size required for singletons to explain this proportion of heritability reflects the low variance contribution of rare variants as a consequence of their low allele frequency (2 pq). “${\beta }^2$ independent of p” shows estimates of the median proportion of heritability explained by singletons if effect size and allele frequency were independent (i.e., singletons and common variants had equal effects on gene expression).