Prediction of pharmacokinetic parameters for diverse drug compoundsтезисы доклада

Работа с тезисами доклада

[1] Prediction of pharmacokinetic parameters for diverse drug compounds / E. V. Radchenko, A. S. Dyabina, V. A. Palyulin, N. S. Zefirov // 19th EuroQSAR Knowledge Enabled Ligand Design. — Vienna, Austria, 2012. The absorption, distribution, metabolism and excretion (ADME) of drugs and drug-like compounds control their bioavailability and play an important role in determining their activity. Substantial work has been devoted to the modelling and prediction of these properties. However, the applicability and usefulness of available models is often diminished due to limited data sets, inaccurate data, and/or insufficiently validated modelling approaches. We have attempted to build generally applicable predictive models for a number of important ADME parameters such as blood-brain barrier permeability (LogBB) and human intestinal absorption (%HIA). The fragmental descriptors of up to 6 atoms were used in the conjunction with back-propagation neural networks (BPNN). The descriptor subset for BPNN modelling was preselected using fast stepwise multiple linear regression (FSMLR). The model predictivity was assessed by 5x4-fold double cross-validation procedure. For the blood-brain barrier permeability (LogBB) we have compiled, to our knowledge, the most complete data set based on the published data. More than a hundred additional compounds were included and the values were verified (and errors corrected) against original publications. On the other hand, inorganic and small organic molecules irrelevant to medicinal chemistry were excluded. The final dataset contained 510 diverse organic compounds. The optimal model has Q2 value (double cross-validation) of 0.79 and RMSE of 0.34. The comparison of the experimental and predicted values is shown in the figure. For the human intestinal absorption (%HIA) we have compiled an extended and corrected dataset of 708 diverse organic compounds. Unfortunately, the distribution of values is far from uniform, with more than 60% of compounds having %HIA > 80. The optimal model has Q2 = 0.71 and RMSE = 16.6. However, for a small fraction of compounds the prediction errors were rather high. After exclusion of 31 outliers to improve the model stability, the final model was obtained that has Q2 = 0.80 and RMSE = 13.5. The constructed models were implemented in a convenient predictor software.

Публикация в формате сохранить в файл сохранить в файл сохранить в файл сохранить в файл сохранить в файл сохранить в файл скрыть