TABLE 3.
Tissue specificity features.
| Feature - tissue specificity | Dominant prediction | RIG a | Support (%) | Lift 1 b | Lift 0 b |
|---|---|---|---|---|---|
| Contains “muscle” | 1 | 0.012 | 13.22 | 1.19 | 0.55 |
| Contains “heart” | 1 | 0.011 | 13.96 | 1.18 | 0.57 |
| Contains “brain” | 1 | 0.008 | 17.48 | 1.14 | 0.67 |
| Contains “testis” | 0 | 0.007 | 0.92 | 0.33 | 2.57 |
a
Relative information gain.
b
Lift indicates the effect fold for the feature on the prediction of being (Lift 1) and not being (Lift 0) miRNA, regulated.
They emphasize which class (1 - regulated, 0 - not regulated by miRNA) the feature is strongest forthe lift values refer to how much more likely the class is, given that feature. e.g., 1.2× for class 0 = class 0 is 20% more likely when that feature is true.