Table 4.
Set | Gene count | No. of UR synonymous | No. of UR LGD mutations | Expected UR LGD mutations | Class vulnerability | Z-score |
Rec. DN LGD in aut | 39 | 2,568 | 79 | 181 | 0.4 | 3.2 |
DN LGD in sib | 173 | 7,372 | 881 | 694 | 1.3 | −2.7 |
FMRP | 795 | 46,230 | 1,521 | 3,681 | 0.4 | 15.7 |
Chromatin | 408 | 14,299 | 807 | 1,467 | 0.6 | 8.2 |
Embryonic | 1,865 | 52,261 | 4,673 | 5,990 | 0.8 | 8.6 |
PSD | 1,398 | 43,183 | 2,638 | 4,607 | 0.6 | 13.9 |
Essential | 1,732 | 52,243 | 3,571 | 5,769 | 0.6 | 13.2 |
UR LGD mutational loads are shown for seven classes of genes: the gene targets of recurrent DN LGD mutations in children with autism (Rec. DN LGD in aut), the targets of DN LGD mutations in unaffected siblings (DN LGD in sib), FMRP-associated genes, the genes encoding chromatin modifiers, embryonic genes, the genes encoding postsynaptic density proteins (PSD), and essential genes (10). For each gene class, we list the observed number of UR synonymous and LGD variants, the expected number of UR LGD variants, and class vulnerability (the ratio of the observed to expected UR LGD variants). Expectations are computed with a simple permutation approach that addresses the nonlinear dependence of the UR LGD variants to gene length. We perform 10,000 random permutations in which each gene in the class is replaced with a random gene with the same number of UR synonymous mutations, and in each permutation, we record the number of UR LGD mutations in the randomly selected genes. We take the mean of the random UR LGD class loads as an expected number of UR LGD mutations for the gene class. We then use the SD of the 10,000 random UR LGD class loads to compute a Z-score as the number of SDs separating the observed and expected class loads, with positive Z-scores when the observed is smaller than the expected and negative otherwise. With the exception of the targets of DN mutation in unaffected children, all classes have a significantly decreased UR LGD mutational load.