A carregar...
Publicação
An In Silico protocol to evaluate and optimize Cyclodextrins for drug delivery
| Nome: | Descrição: | Tamanho: | Formato: | |
|---|---|---|---|---|
| 94.75 MB | Adobe PDF |
Autores
Resumo(s)
Pharmacokinetics plays a significant role in determining the pharmacological effect of drugs, and exploring delivery systems that can enhance these properties is crucial for drug development. Cyclodextrins (CDs) are a group of cyclic oligosaccharides characterized by a hydrophilic outer surface and a lipophilic central cavity that can form inclusion complexes with various drugs, often enhancing their aqueous solubility and making them suitable vehicles for drug delivery. In this work, we highlight how Molecular Docking calculations, Molecular Dynamics (MD) simulations, Quantum Mechanical (QM) calculations, and Machine Learning (ML) can be used to study the interactions between drugs and CDs. Three cyclodextrin systems (HPαCD, HPβCD, and HPγCD) were studied in combination with 39 different drugs. From the Molecular Docking calculations, a tweaked version of the Vina scoring function with a hydrophobic weight of −0.3159 was used, yielding the best correlation with the experimental data, with a Pearson Correlation coefficient (r) of 0.575. The Molecular Dynamics simulations enabled the evaluation of the systems over time, and several analyses were performed, including interface area, and energy landscapes, all of which allowed for a description of the drug’s behaviors. Subsequently, Molecular Mechanics/Poisson-Boltzmann Surface Area (MM/PBSA) calculations were performed, and a problem with the solvation description of this method was identified. The QM calculations were performed with the combination of the ωB97X-D functional and the 6-31G* basis set with a single point refinement of the energy using the def2-TZVP basis set and a quasi-harmonic correction to the entropy. This method was identified as too time-consuming for these many systems. Then, ML models were created, the best one being developed from a set of external data and using the Random Forest regressor/ classifier algorithm. The regression model exhibits poor performance, with an r of 0.40; however, an accuracy of 72.2% for the classifier model was obtained.
Descrição
Tese de Mestrado, Química, 2025, Universidade de Lisboa, Faculdade de Ciências
Palavras-chave
Cyclodextrins Molecular Docking Molecular Dynamics (MD) Quantum Mechanics (QM) Machine Learning (ML)