| Abstract [eng] |
The main objective of this study was the development of QSAR models for drug metabolism-related properties. Novel GALAS (Global, Adjusted Locally According to Similarity) modeling method was used, which is a combination of baseline global QSAR model and local similarity based corrections. GALAS modeling method allows forecasting the reliability of prediction thus defining the model applicability domain. Models predicting CYP3A4 inhibition and regioselectivity of metabolism in human liver microsomes were developed and validated using external test sets. In all cases the baseline models already showed acceptable results, and the overall accuracy of predictions increased after the similarity based corrections. Moreover, the numbers of mispredictions reduced significantly when only results of higher reliability were taken into account. However, the original models are applicable only for less than a half of external datasets. Since the similarity correction procedure of GALAS modeling method allows simple model training, the possibility to expand the applicability domain has been tested. The CYP3A4 inhibition model was successfully adapted to PubChem data and compounds with a novel chemical scaffold. After training the regioselectivity model new metabolism sites could be identified in compounds of new chemical class. Moreover, this model was adapted for human cytochrome P450 isoform profiling. |
