Refining the role of de novo protein-truncating variants in neurodevelopmental disorders by using population reference samples

Abstract

Recent research has uncovered an important role for de novo variation in neurodevelopmental disorders. Using aggregated data from 9,246 families with autism spectrum disorder, intellectual disability, or developmental delay, we found that ∼1/3 of de novo variants are independently present as standing variation in the Exome Aggregation Consortium's cohort of 60,706 adults, and these de novo variants do not contribute to neurodevelopmental risk. We further used a loss-of-function (LoF)-intolerance metric, pLI, to identify a subset of LoF-intolerant genes containing the observed signal of associated de novo protein-truncating variants (PTVs) in neurodevelopmental disorders. LoF-intolerant genes also carry a modest excess of inherited PTVs, although the strongest de novo-affected genes contribute little to this excess, thus suggesting that the excess of inherited risk resides in lower-penetrant genes. These findings illustrate the importance of population-based reference cohorts for the interpretation of candidate pathogenic variants, even for analyses of complex diseases and de novo variation.

Figure 1

Illustration of class 1 and class 2 de novo variants with the genotypes of each variant for 8 of the 60,706 individuals in ExAC.

Figure 2

Properties of class 2 de novo variants. (a) The proportion of de novo variants across each cohort split between class 1 (left) and class 2 (right) with CpG variants marked in black. Class 2 de novo variants are strongly enriched for CpG variants (P < 10⁻²⁰). The corresponding figure using the non-psychiatric version of ExAC can be found in Supplementary Figure 2. (b) Expected recurrence rate (rate of class 2 de novo variants across ID/DD, ASD, and unaffected ASD siblings) given the sample size of the reference dataset. The red dot indicates the observed recurrence rate of the non-psychiatric version of ExAC. (c) Allele frequency distribution of class 2 de novo CpGs by cohort with the matching distribution of CpG variants in ExAC for comparison. Allele frequency distributions do not significantly differ (P=0.14; Wilcoxon rank sum test). Error bars represent 95% confidence intervals throughout (a) – (c). ID/DD, intellectual disability / developmental delay; ASD, autism spectrum disorder.

Figure 2

Properties of class 2 de novo variants. (a) The proportion of de novo variants across each cohort split between class 1 (left) and class 2 (right) with CpG variants marked in black. Class 2 de novo variants are strongly enriched for CpG variants (P < 10⁻²⁰). The corresponding figure using the non-psychiatric version of ExAC can be found in Supplementary Figure 2. (b) Expected recurrence rate (rate of class 2 de novo variants across ID/DD, ASD, and unaffected ASD siblings) given the sample size of the reference dataset. The red dot indicates the observed recurrence rate of the non-psychiatric version of ExAC. (c) Allele frequency distribution of class 2 de novo CpGs by cohort with the matching distribution of CpG variants in ExAC for comparison. Allele frequency distributions do not significantly differ (P=0.14; Wilcoxon rank sum test). Error bars represent 95% confidence intervals throughout (a) – (c). ID/DD, intellectual disability / developmental delay; ASD, autism spectrum disorder.

Figure 2

Properties of class 2 de novo variants. (a) The proportion of de novo variants across each cohort split between class 1 (left) and class 2 (right) with CpG variants marked in black. Class 2 de novo variants are strongly enriched for CpG variants (P < 10⁻²⁰). The corresponding figure using the non-psychiatric version of ExAC can be found in Supplementary Figure 2. (b) Expected recurrence rate (rate of class 2 de novo variants across ID/DD, ASD, and unaffected ASD siblings) given the sample size of the reference dataset. The red dot indicates the observed recurrence rate of the non-psychiatric version of ExAC. (c) Allele frequency distribution of class 2 de novo CpGs by cohort with the matching distribution of CpG variants in ExAC for comparison. Allele frequency distributions do not significantly differ (P=0.14; Wilcoxon rank sum test). Error bars represent 95% confidence intervals throughout (a) – (c). ID/DD, intellectual disability / developmental delay; ASD, autism spectrum disorder.

Figure 3

Partitioning the rate of de novo variants per exome by class 1, class 2, and pLI. Within each grouping, the rate – variants per individual – is shown for ID/DD (left), ASD (middle), and unaffected ASD siblings (right) with the number of individuals labeled in the legends. (a) Rate of de novo synonymous variants per exome partitioned into class 2 (middle) and class 1 (right). No significant difference was observed for any grouping of de novo synonymous variants. (b) Rate of de novo PTVs per exome partitioned into class 2 (middle) and class 1 (right). Only class 1 de novo PTVs in ID/DD and ASD show association when compared to unaffected ASD siblings. (c) Rate of class 2 de novo PTVs broken by different ExAC global allele frequency (AF) thresholds: singleton (observed once; left), AF < 0.0001 (middle), and AF < 0.001 (right). (d) Rate of class 1 de novo PTVs partitioned into class 1 de novo PTVs in pLI ≥ 0.9 genes (right), and class 1 de novo PTVs in pLI < 0.9 genes (middle). The entire observed association for de novo PTVs resides in class 1 de novo PTVs in pLI ≥ 0.9 genes. For all such analyses, the rate ratio and significance were calculated by comparing the rate for ID/DD and ASD to the rate in unaffected ASD siblings using a two-sided Poisson exact test for synonymous variants and one-sided for the remainder (Online Methods). Error bars represent 95% confidence intervals throughout (a) – (d). See Supplementary Fig. 3 for the corresponding figures using the non-psychiatric version of ExAC. ID/DD, intellectual disability / developmental delay; ASD, autism spectrum disorder; PTV, protein truncating variant; pLI, probability of loss-of-function intolerance; NS, not significant.

Figure 3

Partitioning the rate of de novo variants per exome by class 1, class 2, and pLI. Within each grouping, the rate – variants per individual – is shown for ID/DD (left), ASD (middle), and unaffected ASD siblings (right) with the number of individuals labeled in the legends. (a) Rate of de novo synonymous variants per exome partitioned into class 2 (middle) and class 1 (right). No significant difference was observed for any grouping of de novo synonymous variants. (b) Rate of de novo PTVs per exome partitioned into class 2 (middle) and class 1 (right). Only class 1 de novo PTVs in ID/DD and ASD show association when compared to unaffected ASD siblings. (c) Rate of class 2 de novo PTVs broken by different ExAC global allele frequency (AF) thresholds: singleton (observed once; left), AF < 0.0001 (middle), and AF < 0.001 (right). (d) Rate of class 1 de novo PTVs partitioned into class 1 de novo PTVs in pLI ≥ 0.9 genes (right), and class 1 de novo PTVs in pLI < 0.9 genes (middle). The entire observed association for de novo PTVs resides in class 1 de novo PTVs in pLI ≥ 0.9 genes. For all such analyses, the rate ratio and significance were calculated by comparing the rate for ID/DD and ASD to the rate in unaffected ASD siblings using a two-sided Poisson exact test for synonymous variants and one-sided for the remainder (Online Methods). Error bars represent 95% confidence intervals throughout (a) – (d). See Supplementary Fig. 3 for the corresponding figures using the non-psychiatric version of ExAC. ID/DD, intellectual disability / developmental delay; ASD, autism spectrum disorder; PTV, protein truncating variant; pLI, probability of loss-of-function intolerance; NS, not significant.

Figure 3

Partitioning the rate of de novo variants per exome by class 1, class 2, and pLI. Within each grouping, the rate – variants per individual – is shown for ID/DD (left), ASD (middle), and unaffected ASD siblings (right) with the number of individuals labeled in the legends. (a) Rate of de novo synonymous variants per exome partitioned into class 2 (middle) and class 1 (right). No significant difference was observed for any grouping of de novo synonymous variants. (b) Rate of de novo PTVs per exome partitioned into class 2 (middle) and class 1 (right). Only class 1 de novo PTVs in ID/DD and ASD show association when compared to unaffected ASD siblings. (c) Rate of class 2 de novo PTVs broken by different ExAC global allele frequency (AF) thresholds: singleton (observed once; left), AF < 0.0001 (middle), and AF < 0.001 (right). (d) Rate of class 1 de novo PTVs partitioned into class 1 de novo PTVs in pLI ≥ 0.9 genes (right), and class 1 de novo PTVs in pLI < 0.9 genes (middle). The entire observed association for de novo PTVs resides in class 1 de novo PTVs in pLI ≥ 0.9 genes. For all such analyses, the rate ratio and significance were calculated by comparing the rate for ID/DD and ASD to the rate in unaffected ASD siblings using a two-sided Poisson exact test for synonymous variants and one-sided for the remainder (Online Methods). Error bars represent 95% confidence intervals throughout (a) – (d). See Supplementary Fig. 3 for the corresponding figures using the non-psychiatric version of ExAC. ID/DD, intellectual disability / developmental delay; ASD, autism spectrum disorder; PTV, protein truncating variant; pLI, probability of loss-of-function intolerance; NS, not significant.

Figure 3

Partitioning the rate of de novo variants per exome by class 1, class 2, and pLI. Within each grouping, the rate – variants per individual – is shown for ID/DD (left), ASD (middle), and unaffected ASD siblings (right) with the number of individuals labeled in the legends. (a) Rate of de novo synonymous variants per exome partitioned into class 2 (middle) and class 1 (right). No significant difference was observed for any grouping of de novo synonymous variants. (b) Rate of de novo PTVs per exome partitioned into class 2 (middle) and class 1 (right). Only class 1 de novo PTVs in ID/DD and ASD show association when compared to unaffected ASD siblings. (c) Rate of class 2 de novo PTVs broken by different ExAC global allele frequency (AF) thresholds: singleton (observed once; left), AF < 0.0001 (middle), and AF < 0.001 (right). (d) Rate of class 1 de novo PTVs partitioned into class 1 de novo PTVs in pLI ≥ 0.9 genes (right), and class 1 de novo PTVs in pLI < 0.9 genes (middle). The entire observed association for de novo PTVs resides in class 1 de novo PTVs in pLI ≥ 0.9 genes. For all such analyses, the rate ratio and significance were calculated by comparing the rate for ID/DD and ASD to the rate in unaffected ASD siblings using a two-sided Poisson exact test for synonymous variants and one-sided for the remainder (Online Methods). Error bars represent 95% confidence intervals throughout (a) – (d). See Supplementary Fig. 3 for the corresponding figures using the non-psychiatric version of ExAC. ID/DD, intellectual disability / developmental delay; ASD, autism spectrum disorder; PTV, protein truncating variant; pLI, probability of loss-of-function intolerance; NS, not significant.

Figure 4

Phenotypic associations for ASD de novo PTVs. (a) IQ distribution of class 1 de novo PTVs in pLI ≥ 0.9 genes (red) and remaining de novo PTVs (class 2 and class 1 pLI < 0.9; grey) in 393 individuals with ASD with a measured full-scale IQ. Dots indicate the rate in unaffected ASD siblings for their respective categories of de novo PTVs. P-values calculated using Poisson regression. Only class 1 de novo PTVs show association with full-scale IQ. (b) Rate of class 1 de novo PTVs (left set) and the remaining de novo PTVs (class 2 & class 1 in LoF-tolerant genes, right set) in ID/DD (left two bars) and ASD (middle four bars) split by sex and ID with the number of individuals labeled in the legends. Error bars represent 95% confidence intervals, and P-values were calculated using one-side Poisson exact tests comparing to unaffected ASD siblings. ASD, autism spectrum disorder; PTV, protein truncating variant; pLI, probability of loss-of-function intolerance.

Figure 4

Phenotypic associations for ASD de novo PTVs. (a) IQ distribution of class 1 de novo PTVs in pLI ≥ 0.9 genes (red) and remaining de novo PTVs (class 2 and class 1 pLI < 0.9; grey) in 393 individuals with ASD with a measured full-scale IQ. Dots indicate the rate in unaffected ASD siblings for their respective categories of de novo PTVs. P-values calculated using Poisson regression. Only class 1 de novo PTVs show association with full-scale IQ. (b) Rate of class 1 de novo PTVs (left set) and the remaining de novo PTVs (class 2 & class 1 in LoF-tolerant genes, right set) in ID/DD (left two bars) and ASD (middle four bars) split by sex and ID with the number of individuals labeled in the legends. Error bars represent 95% confidence intervals, and P-values were calculated using one-side Poisson exact tests comparing to unaffected ASD siblings. ASD, autism spectrum disorder; PTV, protein truncating variant; pLI, probability of loss-of-function intolerance.