Fig. 2From: Clinical analysis of germline copy number variation in DMD using a non-conjugate hierarchical Bayesian modelError in multivariate normal hyperparameter estimation Original parameter (mean and covariance) values were derived from representative estimates for 79 targets across DMD (and an additional baseline target) using a cohort of high coverage samples. Each point represents the mean absolute percent error across 5 simulated sets of subjects at the coverage and cohort size indicated. For example, this is calculated as follows for μ and values of estimated \(\hat {\mu }\): \(100 * \frac {1}{5} {\sum \nolimits }_{k=1}^{5} \frac {1}{79}{\sum \nolimits }_{i} \vline \frac {\mu _{i} - \hat {\mu _{i}} }{ \mu _{i}} \vline \)a shows percent error averaged across μ; b shows percent error averaged across the expected normalized xi values; c mean and d median percent error across terms in Σ. Legend values indicate total read pair counts (including baseline targets) for each coverage level simulatedBack to article page