RF-Phos: A Novel General Phosphorylation Site Prediction Tool Based on Random Forest

Figure 2

Feature distribution. (a) Distribution of the feature importance of all 593 features for Ser (top), Thr (middle), and Tyr (bottom). Features and corresponding indices are noted. Dashed lines represent boundaries between feature indices. (b) Top ten important features Ser (top), Thr (middle), and Tyr (bottom). The bar labels indicate the feature type to which the important features belong. CTD: composition, transition, and distribution; ASA: accessible surface area; SF: sequence features; ACH: average cumulative hydrophobicity; OP: overlapping properties.

(a)

(b)