Fig 1.
Observed and expected PTVs in the study population.
A: Fraction of genes with at least one stop-gain or frameshift variant as a function of the number of sampled PTVs. The gray curve shows the expected number of genes under a model of neutral de novo mutation rate [12] representing the null hypothesis (no deleterious effects). The green curve shows the number of genes observed with at least one PTV. The orange curve limits the number of observed genes to those hosting highly damaging variants [13]. The purple curve shows the predicted number of genes with at least one PTV under the estimated best-fit parameters under model A–bootstrap replicas of this fit is shown by pale gray (see Methods). B: Extrapolation of the observed number of genes with at least one PTV assuming a model that includes the possibility of finding PTVs due to biological and technical noise. The purple curve shows the predicted number of genes with at least one PTV under the estimated best-fit parameters, while the green curve shows the observed data. Decomposition of the observed and predicted number of genes with at least one PTV: variants in non-haploinsufficient genes (blue) saturate early; variants found in haploinsufficient genes (red) continue to accumulate PTVs due to the constant contribution of biological and technical noise.
Table 1.
Characteristics of the subset of genes (n = 4,204) observed without PTVs after sequencing 16,260 protein coding autosomal genes in 11,546 individuals.
Tests compare genes with and without heterozygous PTVs.
Fig 2.
Expected and observed number of PTVs per gene.
Each dot in the scatter plot corresponds to a gene. X-axis reflects the expected number of PTVs for each gene according to a model of neutral variation based on synonymous variants (Model B, see Methods) while on Y-axis indicates observed number of PTVs in the study dataset. Genes intolerant to heterozygous PTVs with a posterior probability of ≥ 0.99 are colored in red. The distribution shows that there is a continuum of intolerance to PTVs with a general paucity of observed versus expected truncations in the coding genome. The gray line has a slope of 1.