1 / 14

Predicting Blood-Brain Permeation from Three-Dimensional Molecular Structure

Predicting Blood-Brain Permeation from Three-Dimensional Molecular Structure. Patrizia Crivori, Gabriele Cruciani, Pierre-Alain Carrupt, and Bernard Testa. J. Med. Chem 2000 43: 2204-2216 Presented by Ankit Garg. To cross or not to cross?.

dacia
Download Presentation

Predicting Blood-Brain Permeation from Three-Dimensional Molecular Structure

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript

  1. Predicting Blood-Brain Permeation from Three-Dimensional Molecular Structure Patrizia Crivori, Gabriele Cruciani, Pierre-Alain Carrupt, and Bernard Testa. J. Med. Chem 2000 43: 2204-2216 Presented by Ankit Garg

  2. To cross or not to cross? • Some drugs must cross the blood-brain barrier (BBB), others absolutely should not. • Experimentally screening for BBB permeability is expensive.

  3. Many factors influence BBB permeability • H-bonding capacity • Hydrophobicity • Ionization profile • Molecular size • Lipophilicity • Flexibility • Plasma protein binding • Active efflux from CNS • Metabolism

  4. The VolSurf difference • Past approaches have emphasized one or a few factors, without regard to the rest: • Lipophilicity • Solvatochromatic parameters • Topological indices • VolSurf uses 72 descriptors derived from 3D molecular interaction fields.

  5. Multivariate Analysis: PCA • Represent multivariate data along a few, principal component axes. Output Plots Multivariate Data

  6. Two Sets of Data: • “Training” set was based on 44 compounds used in prior work. • “External Prediction” set was based on 108 wide-ranging drugs from literature. • Compounds in both sets had well-characterized BBB behavior.

  7. Choosing Descriptors • VolSurf descriptors are obtained directly from 3D molecular interaction fields. Not sure what this means. • Main difference appears to be that VolSurf descriptors have a clear chemical meaning.

  8. Encouraging 1st Results! PCA on the training set

  9. PCA on BBB+ compounds of the second data set 90% Accuracy! (40 out of 44)

  10. PCA on BBB- compounds of the second data set 65% Accuracy! (46 out of 71)

  11. Reasons for differences in accuracy • Many BBB- compounds passively defuse into the brain but: • are then metabolized before acting • are actively effluxed • At least one compound, Mequitazine, is likely misclassified in the literature based on the results of this study and another one as well.

  12. PLS discriminant analysis: Finding two latent variables Unlike PCA, relies on training, so descriptors can be differentially biased >90% accurate, though with a confidence interval

  13. Most important descriptors Though some descriptors are clearly more important, in general a balance of many descriptors control BBB permeability

  14. Questions: • Paper notes that it is impossible to model mixtures of stereoisomers in 3D? Why is this the case? • To account for conformational variety in the tested molecules, the authors used two different programs to generate low-energy conformations for the same molecule. The data they then generated for these pair complements was similar, suggesting that the conformational differences mattered little. Is this analysis convincing? • What are the relative strengths and weaknesses of using PCA vs. PLS discriminant analysis?

More Related