Combining machine learning, molecular dynamics, and free energy analysis for (5HT)-2A receptor modulator...

Area of Science:

• Computational chemistry and pharmacology
• Artificial intelligence in drug discovery
• Neuroscience and medicinal chemistry

Background:

• The 5-Hydroxytryptamine (5HT)-2A receptor is a critical target for psychoactive drug development.
• Designing selective compounds for the 5HT-2A receptor presents significant challenges.
• Understanding the mechanisms of action for 5HT-2A modulators is crucial for effective drug design.

Purpose of the Study:

• To develop and validate machine learning models for classifying bioactivity mechanisms against the 5HT-2A receptor.
• To integrate machine learning with molecular modeling to enhance the understanding of drug interactions.
• To introduce a novel metric for evaluating drug efficacy against the 5HT-2A target.

Main Methods:

• Construction and evaluation of neural network and XGBoost machine learning models.
• Application of molecular modeling techniques, including molecular dynamics simulations.
• Binding free energy analysis to elucidate drug-receptor interactions.
• Development of a specific 'binding free energy fingerprint' for 5HT-2A modulators.

Main Results:

• Achieved an overall accuracy of approximately 87% in classifying bioactivity mechanisms using ML models.
• Enhanced model performance and gained insights into mechanisms of direct modulators and prodrugs through ML-MM integration.
• Successfully developed a novel 'binding free energy fingerprint' for 5HT-2A modulators.

Conclusions:

• The integration of AI and structural biology provides a powerful workflow for advancing psychoactive drug discovery.
• The developed 'binding free energy fingerprint' offers a new metric for assessing drug efficacy.
• This approach holds significant promise for the rational design of selective 5HT-2A receptor modulators.