Figure 4.

Evaluation of linear regression models made only of known functional polymorphisms. The -log₁₀ P value for the fit of the models applied to each data set are represented in ordinates and compared with the proper best models (plus signs). Models including only one of the functional polymorphisms are indicated by filled symbols, and models combining two polymorphisms are shown as open symbols. Models included either rs10954213 (filled circles), rs2004640 (filled squares), the CGGGG insertion-deletion polymorphism (indel) (filled triangles), rs10954213 and rs2004640 (open circles), rs10954213 and the CGGGG indel (open squares) or the rs2004640 and the CGGGG indel (open triangles). Comparisons with the proper best model for each data set were either nonsignificantly inferior (n.s.) or inferior with *P < 0.05, **P < 0.01, ***P < 0.001 or ****P < 0.0001. Expression data sets in the X-axis are as in Table 2.