Machine Learning Tool for New Selective Serotonin and Serotonin-Norepinephrine Reuptake Inhibitors

Area of Science:

• Medicinal Chemistry
• Computational Chemistry
• Pharmacology

Background:

• Depression affects approximately 5% of the population, with selective serotonin reuptake inhibitors (SSRIs) and serotonin-norepinephrine reuptake inhibitors (SNRIs) as first-line treatments.
• Developing novel antidepressants requires predicting the affinity and inhibition potential of new chemical entities for serotonin (SERT) and norepinephrine (NET) transporters.

Purpose of the Study:

• To develop robust Quantitative Structure-Activity Relationship (QSAR) models for SERT and NET transporters.
• To predict the affinity (pIC50) and inhibition potential (pKi) of novel molecules targeting SERT and NET.

Main Methods:

• QSAR models were constructed using the Automated Machine Learning tool, Mljar.
• Two-dimensional Mordred descriptors were employed for molecular representation.
• Models were trained on 80% of the data with 10-fold cross-validation and externally validated on the remaining 20%.

Main Results:

• Validated QSAR models demonstrated predictive capabilities for both SERT and NET transporters.
• For NET, R-squared values reached 0.640 (pIC50) and 0.709 (pKi).
• For SERT, R-squared values reached 0.678 (pIC50) and 0.828 (pKi).

Conclusions:

• The developed QSAR models provide reliable predictions for SERT and NET transporter interactions.
• These models are integrated into the SerotoninAI application, facilitating drug discovery for depression.