| Article ID | Journal | Published Year | Pages | File Type |
|---|---|---|---|---|
| 1372424 | Bioorganic & Medicinal Chemistry Letters | 2010 | 5 Pages |
A multiple linear regression QSAR model was developed based on a set of 61 compounds with internally consistent permeability data measured across Franz cell. The data was normalized using a mean permeability value of a reference compound, 3-isobutyl-1-methylxanthine (IBMX). The QSAR model contained only five simple descriptors and had a correlation coefficient, r2 of 0.77 between experimental and calculated values for skin permeability. The mean absolute error (MAE) was 0.3 for the entire set and the cross validation coefficient, q2 was 0.71. The in silico skin permeability model was used as a filter for virtual libraries and to optimize skin permeation of specific compounds for several dermatology discovery projects.
Graphical abstractA multiple linear regression QSAR model was developed based on a set of 61 compounds with internally consistent permeability data measured across Franz cell. The data was normalized using a mean permeability value of a reference compound, 3-isobutyl-1-methylxanthine (IBMX). The QSAR model contained only five simple descriptors and had a correlation coefficient, r2 of 0.77 between experimental and calculated values for skin permeability. The mean absolute error was 0.3 for the entire set and the cross validation coefficient, q2 was 0.71. The in silico skin permeability model was used as a filter for virtual libraries and to optimize skin permeation of specific compounds for several dermatology discovery projects. logPe=6.6177-0.0022∗PISA-0.3516∗donorHB-0.2451∗accptHB-5.8373∗glob+0.2700∗EA(eV)Figure optionsDownload full-size imageDownload as PowerPoint slide
