Computational Model To Predict the Fraction of Unbound Drug in the Brain

Area of Science:

• Pharmacokinetics and Drug Metabolism
• Computational Chemistry
• Neuroscience

Background:

• The unbound drug fraction in the brain (fu,brain) is critical for assessing central nervous system (CNS) drug efficacy and toxicity.
• Currently, no publicly accessible computational model exists to predict fu,brain without experimental measurements.

Purpose of the Study:

• To develop and validate an in silico model for predicting the unbound drug fraction in the brain (fu,brain).
• To provide a freely available tool for researchers to estimate CNS drug behavior.

Main Methods:

• Collected 253 fu,brain measurements from literature and databases.
• Developed in silico predictive models using freely available software.
• Selected optimal chemical descriptors, trained, and rigorously evaluated model performance.

Main Results:

• The developed in silico model achieved acceptable prediction performance on a test dataset.
• Model performance metrics include R² = 0.630 and 69.4% of compounds predicted within a 3-fold error.
• Prediction accuracy was based solely on chemical structure information.

Conclusions:

• The study successfully created a computational model to predict fu,brain.
• This model can aid in the early assessment of CNS drug candidates, reducing experimental burden.
• The model and data are accessible, promoting further research in CNS drug discovery.