Selecting Negative Samples for PPI Prediction Using Hierarchical Clustering Methodology

Figure 5

Comparison of accuracy obtained in negative datasets for the two trained models: the SVM model trained using the training set formed by the GSP set and the GSN set obtained using the proposed hierarchical clustering method (clustered) and the SVM model trained using the training set where the GSN set was randomly selected (Rand. RBF-SVM) and the balanced RBF-SVM is the SVM model trained using the training set formed by the GSP set and the GSN set obtained using the approach to create a “balanced” negative. Please note that Rtest_1, 𝑅^{𝑡𝑒𝑠𝑡}_2, 𝑅^{𝑡𝑒𝑠𝑡}_3, 𝑅^{𝑡𝑒𝑠𝑡}_4, 𝑅^{𝑡𝑒𝑠𝑡}_5, 𝑅^{𝑡𝑒𝑠𝑡}_6, 𝑅^{𝑡𝑒𝑠𝑡}_7, 𝑅^{𝑡𝑒𝑠𝑡}_8, and 𝑅^{𝑡𝑒𝑠𝑡}_9 correspond to: 𝑅^test_13, 𝑅^test_23, 𝑅^test_32, 𝑅^test_43, 𝑅^test_53, 𝑅^test_63, 𝑅^test_73, 𝑅^test_83, and 𝑅^test_93. And the “balanced” negative set is created using the approach by Yu et al.[38].