Since pharmacokinetics and drug toxicity is major reasons of drug attrition, predicting those parameters is useful to reduce the cost of drug discovery. I developed several methods to predict several important parameters related to drug metabolism and drug toxicity: the interaction of drug and cytochrome P450 that is the most important drug metabolizing enzyme, human hepatic clearance, human intrinsic clearance, and human acute toxicity.
Some of my models tried to correct the difference between human in vivo parameter data and its in vitro experimental values. Structural information of drug could decrease the difference. This algorithm was applied to predict human hepatic clearance, human intrinsic clearance, and human acute toxicity.
In addition, I developed a new input variable which describes the average activity of drug toward the most important five drug metabolizing cytochrome P450 isozymes of structurally similar drugs with a query drug. This algorithm was applied to predict substrates, inhibitor, and inducers of cytochrome P450 isozymes. Results showed that this new information could improve the prediction accuracy of conventional models using only structural information of drug. This new input was applied to predict human intrinsic clearance and human acute toxicity.
My models can be useful at the early-stage drug screening, and then they can predict the possibility of drug attrition from predicted pharamcokinetics related to drug metabolism and drug toxicity. I expect that my research helps to reduce the cost and the time of drug discovery.