Figure 1

An ADTree created using 10-fold CV and bootstrapping. The root node indicates the bias in the data set, i.e., the ratio of positive to negative class examples (disease-associated proteins versus non-disease-associated proteins). The rectangles (decision nodes) contain the feature name. The number in parentheses within each decision node indicates the order in which the rule was found. The amount of node conservation between each of the trees generated in the validation step is indicated by the color of the box (red: ≥ 90%, orange: ≥ 70% (none in this tree), yellow: ≥ 50%, green: ≥ 30%, blue: ≥ 10%, black: ≤ 10% (none in this tree)). Ovals (prediction nodes) contain the value for the weighted vote, where a positive number indicates a prediction for disease-association. The numbers next to the arrows correspond to the threshold for the prediction. If the attribute value is equal to or exceeds this number, the left path is followed; otherwise the prediction follows the right path.