Molecular Design Research On Aurora Kinase Target

It is well known that cancer is a disease with higher mortality, therefore, there is an urgent need to discover new drugs against refractory tumors to cure or alleviate cancer patients'illness and suffering. Aurora kinase family is an important class of serine/threonine kinases, including Aurora kinase A, B, C three family members, which have a controlling role in cell mitosis. It has been reported that over-expression of Aurora kinase has been found in variety cancer cells. This thesis studied on Aurora kinase inhibitors.In the first part of this thesis, classification models of Aurora Kinase inhibitors were built using Self-Organizing Map (SOM) and Support Vector Machine (SVM), respectively. The classification models can discriminate the Aurora Kinase inhibitors to three classes:selective inhibitors of Aurora-A kinase, selective inhibitors of Aurora-B kinase and no selectivity. In this study, 10 descriptors were selected to build models. The accuracy of the models in prediction for the training, test sets are 97.0%,88.9% for SOM,95.5% and 100% for SVM, respectively.In the second part of this thesis, two classification models of 148 Aurora-A Kinase inhibitors were developed to separate active and weakly potent active inhibitors of Aurora-A Kinase. Then the classification models were built using a Kohonen's Self-Organizing Map (SOM) and a Support Vector Machine (SVM) method, respectively. The prediction accuracy of the models for the training and test sets are 96.6% and 90.0% for SOM,93.2% and 93.3% for SVM. The classification models we developed could be used for virtual screening an existing database to find possible new lead compounds with higher activity.In the third part of this thesis, several QSAR (Quantitative Structure Activity Relationships) models for predicting the inhibitory activity of 117 Aurora-A kinase inhibitors were developed. Each molecule was represented by 13 selected 2D and 3D molecular descriptors calculated by the ADRIANA.Code. The whole dataset was split into a training set and a test set based on two different methods, (1) by a random selection; and (2) on the basis of a Kohonen's self-organizing map (SOM). Then the inhibitory activity of 117 Aurora-A kinase inhibitors was predicted using Multilinear Regression (MLR) analysis and Support Vector Machine (SVM) methods, respectively. For the two MLR models and the two SVM models, for the training sets, the correlation coefficients of over 0.90 were obtained; for the test sets, the correlation coefficients of over 0.92 were achieved.