Skip to main content
. 2017 Jan 17;114(5):1087–1092. doi: 10.1073/pnas.1612561114

Table 3.

Predictor importance for the best-fit logistic regression model predicting ASE from genomic features, selected using AIC (AIC = 3,086.9)

Model parameter Coeff. (SE) z value P value OR
gbM −0.67 (0.20) −3.41 <10−3 0.51
πN/πS 0.08 (0.04) 2.25 0.024 1.09
Expression level 0.20 (0.06) 3.31 <0.001 1.22
Promoter polymorphism 0.21 (0.05) 4.45 <10−3 1.23
Tissue specificity 0.30 (0.06) 5.03 <10−3 1.35
TE within 1 kb 0.32 (0.13) 2.50 0.013 1.38
Coexpression module size −0.08 (0.05) −1.59 NS 0.92
Gene length 0.08 (0.06) 1.49 NS 1.09
Intercept −2.60 (0.06) −42.91 <10−3 0.07

Regression coefficients (Coeff.) and their SE, z statistics and associated P values, and odds ratios (OR) are shown.