In this work we analyzed the currently known human protein interaction network and its relationship to disease using the ADTree algorithm.
To evaluate potential markers for PZ and TZ tissues a classification procedure was performed using the ADTree algorithm, which generates trees where each decision node specifies a miRNA and a threshold expression value, while the prediction nodes contain numbers, which are summed up when the classification is done.
The ADtree algorithm was used to identify markers for classification of tissues and a cross-validation procedure was used to test the generality of the identified miRNA-based classifiers.
The ADTree algorithm in the WEKA data mining tool was used to identify zone-specific signatures, in the form of alternating decision trees [ 29, 30], for classification of tissues.
Instead, we compared the performance of the ADTree algorithm on our data set with that of a variety of other tree-based classifiers as well as Bayesian, function-based, and meta-classifiers using the Weka machine learning workbench [ 40]. Figure 2 shows ROC curves for the ADTree, Adaboost [ 46], Bayesian Network, Naïve Bayes, and Radial Basis Function (RBF) Network [ 47] classifiers.
The benefits of the ADTree algorithm are two-fold: first, it provides a confidence score for each example based on its full traversal path through the tree.
Here, we use the ADTree model to discover the rules that distinguish binding and non-binding domains.
A data-mining model generated using the ADTree ensemble technique improved the prediction of AxLN metastasis in patients with primary breast cancer, compared with older models such as the MSKCC nomogram.
Conclusions: The high accuracy of the learned models and good agreement between the rules discovered using the ADtree classifier and mechanisms reported in the literature reflect the value of machine learning protocols in both prediction and biological knowledge discovery.
Seven attributes are filtered out by the ADTree algorithms after screening, including cyanotic lips, cyanotic hands and fingernails, tortuous collaterals, dark blue tongue, dark red lips, and gingival and thickness of collaterals, altogether forming another decision tree model.
