Integrating de novo and inherited variants in 42,607 autism cases identifies mutations in new moderate-risk genes

Abstract

To capture the full spectrum of genetic risk for autism, we performed a two-stage analysis of rare de novo and inherited coding variants in 42,607 autism cases, including 35,130 new cases recruited online by SPARK. We identified 60 genes with exome-wide significance (P < 2.5 × 10−6), including five new risk genes (NAV3, ITSN1, MARK2, SCAF1 and HNRNPUL2). The association of NAV3 with autism risk is primarily driven by rare inherited loss-of-function (LoF) variants, with an estimated relative risk of 4, consistent with moderate effect. Autistic individuals with LoF variants in the four moderate-risk genes (NAV3, ITSN1, SCAF1 and HNRNPUL2; n = 95) have less cognitive impairment than 129 autistic individuals with LoF variants in highly penetrant genes (CHD8, SCN2A, ADNP, FOXP1 and SHANK3) (59% vs 88%, P = 1.9 × 10−6). Power calculations suggest that much larger numbers of autism cases are needed to identify additional moderate-risk genes.

Integration of single cell gene expression data in Bayesian association analysis of rare variants

“…This is in part what motivated the design of VBASS. We obtained ASD DNV data from a recent preprint 15 that combined exome and genome data from four studies (see Methods), and single cell RNA-seq data of human fetal midbrain and prefrontal cortex from two publications 21,22 . We applied VBASS and extTADA to the full ASD data set with 16616 trios.…”

“…Given the severe consequence of these conditions on reproductive fitness, risk variants with large effect are under strong negative selection and therefore have low frequency in the population. Recent genetics studies identified hundreds of risk genes of these conditions, largely by rare de novo variants [2][3][4][5][6][7][8][9][10][11] , however, the majority of risk genes remain unidentified 10,[12][13][14][15] , due to challenges in statistical power in analysis of rare variants 16 .…”

AlphaCluster: Coevolutionary driven residue-residue interaction models enable quantifiable clustering analysis of de novo variants to enhance predictions of pathogenicity

“…Reference source not found. for a detailed overview of the previous evidence of clustering of missense variants in these eight proteins from Kaplanis et al 1 and Zhou et al 9 . Across various clustering tests (1D clustering, 3D clustering and 3D clustering with gMVP score scaling and a threshold of rank score 0.7), we calculated the mean p-values for a given random subsample of the total missense variants of the known gene over 100 trials, obtaining missense variants from random cohort samples of fixed sizes (ranging from 100, 500, 5,000, 10,000, 15,000, 20,000, and the full cohort of 21,020 for the autism cohort, and 100, 500, 5,000, 10,000, 15,000, 20,000, 25,000 and the full cohort of 31,783 for the NDD cohort).…”

“…Several autism cohorts were used as a source of de novo variants from affected probands (SPARK 9,31 , SSC 32 and ASC 33 ), whereas the cohort from Kaplanis et al was used to source NDD de novo variants. Cohort information for autism an NDD is presented in Error!…”

SHINE: Protein Language Model based Pathogenicity Prediction for Inframe Insertion and Deletion Variants

“…Besides, the two datasets were not used to train any of the previous prediction models for indels, so are good benchmarks for method comparison. Six hundred and eighty-six NDD risk genes were collected based on previous studies (21,22). De novo inframe indels in the 686 NDD genes in two NDD cohorts were used as proxies of pathogenic variants, and short inframe indels with at most three-amino-acid deletion/insertion in the UK biobank (200,000 exomes) were considered benign.…”