Machine learning for discrimination of phospholipidosis–inducing drugs
-
- KURODA Yukihiro
- School of Pharmacy and Pharmaceutical Sciences, Mukogawa Women's University
-
- IWAKUMA Yoshie
- School of Pharmacy and Pharmaceutical Sciences, Mukogawa Women's University
Bibliographic Information
- Other Title
-
- 機械学習によるリン脂質症誘発薬の判別方法の検討
Abstract
<p>Phospholipidosis induction potential of cationic drugs roughly correlates with their hydrophobicity. However, it is difficult to discriminate the inducers and non-inducers only by hydrophobicity. Also, correlation between the induction potentials and drug binding to phospholipid membranes have been reported. In this study, we clarified the correlation between the phospholipidosis–induction potentials and the membrane bindings as well as other physicochemical properties. We also tried to discriminate the inducers and non-inducers by machine learning. </p><p>Retention factor to an immobilized artificial membrane column was measured by liquid chromatography. From our previous data, it was suggested that the difference between the predicted reversed–phase retention when the ACN content in the mobile phase was 40% and the experimentally measured value (Δk40) may be used to discriminate between the inducers and non-inducers. In addition to Δk40, Δlogk40 were also determined. Nine other physicochemical parameters including logP, pKa, polar surface area, and logS were used. Twenty-eight cationic drugs were selected for this study. </p><p>The correlation ratios between the induction potentials and the physicochemical parameters were significantly well especially for logP, logS, Δlogk40. Discriminant analysis based on the modified Mahalanobis discriminant function was performed using these parameters as explanatory variables. As a result, correct classifications were obtained, suggesting that machine learning can contribute to clear discrimination of lipidosis-inducers and non-inducers.</p>
Journal
-
- Annual Meeting of the Japanese Society of Toxicology
-
Annual Meeting of the Japanese Society of Toxicology 50.1 (0), P3-313-, 2023
The Japanese Society of Toxicology
- Tweet
Details 詳細情報について
-
- CRID
- 1390017920608094208
-
- Text Lang
- ja
-
- Data Source
-
- JaLC
-
- Abstract License Flag
- Disallowed