Figure 4.

Estimated marginal posterior probabilities for Scenario 1 datasets. The top estimated posterior probabilities were summarized across the Scenario 1 datasets. The box plots confirms that PEAK was converging to the same estimates as the M-H algorithm, regardless of the graph used and the variable mappings. The simulated causal variants X₁₀ and X₇ had elevated posterior probabilities and evidence of association across datasets, while the non-causal variants did not. Note that the whiskers and outliers were not drawn on the box plots.