Figure 7

Summary of DNA methylation data. (A) A plot of the DUX4 gene body estimated average percent methylation for each sample, using a mixture-model to estimate this value for the 4qA allele with the lesser percent. FSHD-affected samples are split into two groups, those with nonmanifesting first-degree relatives in the sample cohort (FSHD(b); orange) and those without (FSHD(a); red). The nonmanifesting samples are labeled NonMfst (green), and the unaffected control samples are labeled Control (grey). Solid symbols indicate myocyte samples, and empty symbols indicate blood samples. Triangles indicate data from the 4qA assay and squares indicate data from the 4qA-L assay. Each group is subdivided into myocyte and blood subgroups. Within each of the eight subgroups, symbols are ordered by family number. Blue crosses behind each subgroup indicate means ± standard errors based on a linear mixed effect (LME) model with fixed effects for each of these eight subgroups, an additive fixed effect for assay type, and a random effect for family. Means and error bars show estimated fixed effects for 4qA assay; 4qA-L estimates are higher. LME calculations were performed on logit-transformed methylation probabilities, and results were then transformed back to percentages using a logistic transformation (which is why the error bars are not symmetric about the means). (B) The same data as in (A), with the same color scheme, but with samples ordered by minimum 4qA EcoRI/BlnI length (ordered by family number in case of ties) and with lines connecting related subjects. Vertical green lines connect FSHD-affected and nonmanifesting pairs (who have the same minimum 4qA EcoRI/BlnI length), and dashed grey lines connect FSHD-affected and control pairs (who do not). Vertical green line is not visible for family 43 (fourth column).