Table 2.
Multiple regression model to predict aberrant splice-site activation and exon skipping
| Predictor variable | Regression coefficient | Standard error | t-value | P-value |
|---|---|---|---|---|
| Exon length | 0.0092 | 0.0017 | 5.36 | <0.0001 |
| SF2/ASF ESE score densitya | 0.0535 | 0.0179 | 2.99 | <0.005 |
| The NN score density (5′ss) | 0.9679 | 0.3258 | 2.97 | <0.005 |
| PESS density | 0.1365 | 0.0673 | 2.03 | <0.05 |
Input data for logistic regression summarized in Table 1 were analysed using a generalized linear model with a logistic link (S-PLUS, v. 7.0).
aThe original SF2/ASF matrix was a consistently preferred variable over the updated (33) matrix in both stepwise and logistic regression analyses. Correlation of the two densities in EXSK and CR-exons was 0.77.