Development of a new predictive model for interactions with human cytochrome P450 2A6 using pharmacophore ensemble/support vector machine (PhE/SVM) approach.

PURPOSE: The objective of this investigation was to yield a generalized in silico model to quantitatively predict CYP2A6-substrates/inhibitors interactions to facilitate drug discovery.
METHODS: The newly invented pharmacophore ensemble/support vector machine (PhE/SVM) scheme was employed to generate the prediction model based on the data compiled from the literature.
RESULTS: The predictions by the PhE/SVM model are in good agreement with the experimental observations for those molecules in the training set (n = 24, r (2) = 0.94, q (2) = 0.85, RMSE = 0.30) and the test set (n = 9, r (2) = 0.96, RMSE = 0.29). In addition, this in silico model performed equally well for those molecules in the external validation sets, namely one set of benzene and naphthalene derivatives (n = 45, r (2) = 0.81, RMSE = 0.46) and one set of amine neurotransmitters (n = 4, r (2) = 0.98, RMSE = 0.32). Furthermore, when compared with crystal structures, the calculated results are consistent with the published CYP2A6-substrate co-complex structure and the plasticity nature of CYP2A6 is also revealed.
CONCLUSIONS: This PhE/SVM model is an accurate and robust model and can be utilized for predicting interactions with CYP2A6, high-throughput screening and data mining to facilitate drug discovery.