The DFE for each disease cohort is obtained by weighting the posterior density of hs for each autosomal gene with the fraction of observed LOF mutations in the gene in that cohort. In each panel, the DFE of all possible LOF mutations is denoted with a black curve. For n de novo mutations (DNMs) in a disease cohort, the gray lines denote 100 bootstrapped DFEs of a set of n DNMs randomly sampled from the full set of LOF mutational opportunities. P-values were calculated from the rank of the mean of the distribution for each disease compared to the means of 1000 bootstrapped distributions (see 'Materials and methods'). The probability of hs >10% for all possible autosomal LOF mutations is 20%. The probability of hs > 10% in each panel denotes the area under the distribution (in red) in the interval (0.1,1) for de novo LOF mutations seen in the corresponding disease cohort. The estimated DFE of observed de novo LOF mutations in individuals affected by (A) developmental disorders, from the Deciphering Developmental Disorders (DDD) cohort (Kaplanis et al., 2020), (B) congenital heart disease (Jin et al., 2017), (C) developmental and epileptic encephalopathy (EuroEPINOMICS-RES Consortium et al., 2014; Hamdan et al., 2017), (D) autism, from the Autism Sequencing Consortium (ASC) and Simons Simplex (SSC) (Satterstrom et al., 2020), (E) schizophrenia (Fromer et al., 2014; Howrigan et al., 2020; Rees et al., 2020; Xu et al., 2012), and (F) Tourette’s syndrome and/or obsessive-compulsive disorder (OCD) (Cappi et al., 2020; Willsey et al., 2017). The estimated DFE of observed de novo LOF mutations in genes (G) recurrently hit in individuals with autism, (H) shared between individuals with autism and schizophrenia, and (I) shared between autism and developmental disorders (DD) or congenital heart disease (CHD).