Genetic model fitting results for variation in liver function test proteins.

The best fitting model is highlighted in grey. For each protein, full models (ACE & ADE) was compared to nested models (AE, CE, E) using a chi-squared test ΔX2 = (X2 sub model)−(X2 full model) with the degrees of freedom equal to ΔDf = (Df sub model)−(Df full model). The degrees of freedom increases from the full to sub or nested models due to drop in the numbers of parameters estimated as one moves down the model hierarchy. To be judged a good-fit, models should have a non-significant chi-squared goodness-of-fit statistic (p>0.05). Note, C and D cannot be included together in the same model as in quantitative genetic studies of human populations they are confounded thus the full model is either ACE or ADE. Comparisons with the ACE full model are shown here. In all cases, ACE provided a better model fit than ADE with a smaller chi-squared goodness-of-fit statistic (data not shown).Abbreviations: X2 = chi-squared goodness-of-fit statistic; Df = degrees of freedom; ΔDf = (df sub model)−(df full model); Δ X2 = (X2 sub model)−(X2 full model); P = P-Value; A = Additive genetic influence; C = Shared environmental variance; E = Unique environmental variance.