Classification Models of Estrogen Receptor-B Ligands Based on PSO-Adaboost-SVM
Zhengjun Cheng, Yuntao Zhang · Journal of Convergence Information Technology · 2010
In this paper, some chemometrics methods have been applied for modeling and predicting classification of Estrogen Receptor-β (ERβ) selective ligands derivatives with radial distribution function (RDF) descriptor calculated from the molecular structure alone for the first time. The particle swarm optimization (PSO) and genetic algorithms (GA) methods have been used to select descriptors which are responsible for the inhibitory activity of these compounds. Mathematical models are obtained by support vector machine (SVM) and Adaptive boosting (Adaboost). The model containing five descriptors founded by Adaboost-SVM has shown better predictive ability than the other models. The total accuracy in prediction for the training and test set are 100.0% and 94.3% for PSO-Adaboost-SVM, 98.0% and 91.4% for PSO-SVM, 100.0% and 88.6% for GA-Adaboost-SVM, 100.0% and 85.7% for GA-SVM, respectively. A comparison with other approaches such as Constitutional, Geometrical, 3D-MoRSE (3D-Molecular Representation of Structure based on Electron diffraction),Topological,GETAWAY(Geometry, Topological and Atoms-Weighted AssemblY) and 2D autocorrelations descriptors has also been carried out.