Table 2.
Predicted vs actual gene expression*.
| Gene | AIC: no random effects | AIC: random effects | Top random effect model | Models in 2 AIC | Intercept | SE | p-value | Slope | SE | p-value (0) | p-value (1) | R2 cond. | R2 marg. |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| IgG | 193.66 | 195.66 | 1 | 0,1 | 0.09 | 0.12 | 0.48 | 0.83 | 0.21 | 0.00 | 0.65 | 0.20 | 0.19 |
| IgM | 214.59 | 213.77 | 2 | 2,3,0 | -0.28 | 0.14 | 0.05 | 0.88 | 0.24 | 0.00 | 0.64 | 0.55 | 0.29 |
| IgA | 187.50 | 187.15 | 1 | 0,1 | -0.09 | 0.14 | 0.53 | 0.67 | 0.19 | 0.00 | 0.09 | 0.32 | 0.18 |
| IgE | 227.66 | 228.45 | 1 | 0,1 | -0.14 | 0.20 | 0.47 | 0.68 | 0.19 | 0.00 | 0.10 | 0.28 | 0.16 |
| CD4 | 163.54 | 165.54 | 1 | 0,1 | -0.08 | 0.13 | 0.56 | 0.80 | 0.17 | 0.00 | 0.23 | NA | 0.26 |
| CD8 | 183.19 | 185.12 | 1 | 0,1 | 0.04 | 0.14 | 0.77 | 0.67 | 0.29 | 0.02 | 0.26 | 0.11 | 0.07 |
| CD11 | 156.39 | 158.39 | 1 | 0,1 | -0.07 | 0.09 | 0.40 | 0.63 | 0.19 | 0.00 | 0.05 | NA | 0.14 |
| CD16 | 151.61 | 153.17 | 1 | 0,1 | 0.03 | 0.10 | 0.77 | 0.77 | 0.15 | 0.00 | 0.13 | 0.35 | 0.29 |
| MHC-II | 223.48 | 225.48 | 1 | 0,1 | -0.31 | 0.24 | 0.20 | 0.40 | 0.34 | 0.25 | 0.08 | NA | 0.02 |
| NKG2D | 209.60 | 210.51 | 1 | 0,1 | -0.24 | 0.18 | 0.18 | 0.65 | 0.25 | 0.01 | 0.17 | 0.20 | 0.09 |
| IgM : IgG | 163.89 | 165.90 | 1 | 0,1 | -0.42 | 0.15 | 0.01 | 0.29 | 0.25 | 0.24 | 0.01 | NA | 0.02 |
| CD4:CD8 | 136.43 | 137.10 | 1 | 0,1 | -0.03 | 0.11 | 0.79 | 1.12 | 0.22 | 0.00 | 0.60 | 0.43 | 0.32 |
*Table shows results for each gene model. Each model was run using four different model types: 0, no random effects, 1, random intercept, 2, random slope, 3, both random intercept and random slope. Akaike Information Criterion (AIC) was reported for model 0, as well as the best fitting random effect model. All models within the top 2 AIC are reported. For the best fitting random effect model, the intercept, slope, standard errors (SE), p-value and R2 values are reported. Each top model was rerun to test if the relationship between the predicted and actual gene expression was significantly different from 1 and 0, with p-values for both analyses reported. Bold indicates significant values.