To compute the Bayes factors and p values, TADA-Denovo requires the following parameters: ntrio: the number parent-child trios passing QC (484)mu: per gene mutation probabilities for each class of de novo variant assessed; in our case, de novo LGD and de novo Mis3 variants. The per gene probability of any mutation, from Sanders et al. (2012), was modified to reflect the chance of each class of mutation by the following steps:The fraction of de novo variants that are LGD was estimated as: theoreticalLGDRatePerChildtheoreticalNonSynRatePerChild+theoretical SynRatePerChildThe fraction of de novo variants that are Mis3 was estimated as: theoreticalMis3RatePerChildtheoreticalNonSynRatePerChild+theoretical SynRatePerChildNext, the per gene overall de novo mutation probability was multiplied by each of these fractions to estimate the per gene probabilities of each class of mutation: dnLGDMu=overallMutationP rob∗fractionLGD dnMis3Mu=overallMutationP rob∗fractionMis3gamma.mean.dn (γ): the average relative risk (γ) is related to the fold-enrichment (λ, relative to random expectation) and the fraction of causal genes (π) by the following equation: π (γ − 1) = λ − 1.We calculated the fraction of causal genes (π) as: 420risk genes17226refseqHg19genes=0.02369(420risk genes were estimated by MLE)Fold-enrichment (λ) for LGD as: numProbandLGDmutationsnumControlLGDmutations∗numberProbandSynMutationsnumberSiblingsSynMutations=3922∗111134=2.14Fold-enrichment (λ) for Mis3 as: numProbandMis3mutationsnumControlMis3mutations∗numberProbandSynMutationsnumberSiblingsSynMutations=160158∗111134=1.22Solving for the relative risk (γ): γ=1+λ−1π=1+2.14−10.02369=49.1155for LGD mutations& γ=1+λ−1π=1+1.22−10.02369=10.3901for Mis3 mutations
