For each Dataset, the average value of the Z-score across all Models (1-5 from Table 2, rounded to the nearest whole number.) Two averages are given |ΔDAF| - Models 2 and 5 before the slash; Models 1, 3 and 4 after the slash. For H|H, a positive number indicates a positive value is a strong predictor for eQTLs, for the other measures, an extreme negative value indicates that a low log p-value is a strong predictor for eQTLs. ΔiHH and iHS are intended to be measures of positive selection. With the exception of |ΔDAF| (which changes substantially depending on whether Fst is included in the model), these Z-scores do not change greatly among Models 1-6 in Table 2, indicating that these predictors are largely independent.