Theoretical Studie On The Predictions Of Inhibitors Of HERG Potassium Channel And Breast Cancer Resistance Protein

The pharmacokinetic properties ?Absorption, Distribution, Metabolism, Excretion, ADME? and Toxicity ?T? of candidate drugs are critical for drug design and development. Therefore, the assessment and optimization of compound's ADMET is essential for the early stages of drug design. Two ADMET-related proteins, hERG channel and breast cancer resistance protein ?BCRP?, were studied in this thesis. The theoretical models for the predictions of inhibitors of hERG channel and BCRP were developed, and the receptor-ligand recognition mechanisms were explored.Blockade of human ether-a-go-go-related gene ?hERG? channel by compounds may lead to drug-induced QT prolongation, arrhythmia and Torsades de Pointes ?TdP?, and therefore reliable prediction of hERG liability in the early stages of drug design is quite important to reduce the risk of cardiotoxicity-related attritions in the later development stages. In the first part of this thesis, we built about 200 pharmacophores based on the data set, and then the important pharmacophores to distinguish hERG blockers and nonblockers were identified by recursive partitioning. Finally, the classifiers based on multiple pharmacophores were built by using Support vector machine ?SVM? and NaiveBayesian Classification ?NBC? techniques. The best SVM model achieved the prediction accuracies of 84.7% for the training set and 82.1% for the test set and external validation set. Notably, the accuracies for the hERG blockers and nonblockers in the test set are 83.6% and 78.2%, respectively. Moreover, the important pharmacophores were highlighted by clustering analysis, which helps to understand the multi-mechanisms of actions of hERG ligands.Multidrug resistance is one of the main reasons for the failure to malignant tumor treatment, and breast cancer resistance protein plays critical role in multidrug resistance. In the second part of this thesis, we collected 860 BCRP blockers and nonblockers. Then, the optimal set of 36 molecular descriptors from a large number of descriptors was determined by simulated annealing and random forest. Based on the optimal descriptors and different molecular fingerprints, the naive Bayesian classification approach was used to develop the classifiers to distinguish BCRP blockers from nonblockers. The best Bayesian classifier based on the optimal descriptors set and LCFP₄ fingerprints achieved the prediction accuracies of 90.1% for the training set,94.2% for the test set and 93.3% for the outer validation set. Moreover, the important structural fragments for BCRP inhibition were discussed.