| Abstract [eng] |
Metabolic stability of a drug-candidate molecule is an essential property in the assessment of the required dosage and potential toxicity to the humans. Ideally, this information should be acquired as early as possible to lower the cost of the drug development process. For these purposes the use of computer models for molecular property prediction is getting increasingly popular. Metabolic stability models commercially available today or published in the literature up to date are usually created using proprietary databases of commercial companies, hence are specific for the chemical space of their research interest. The model described in this work is based on a relatively small yet diverse database, covering a wide chemical space. A notable finding during the model development was the fact that the predicted number of metabolic soft-spots for a compound isn’t descriptive of its metabolic half-life in any way. On the other hand, LogP did have a statistically significant impact on the accuracy of the model predicting molecule’s metabolic stability. |
