Prediction of 5-hydroxytryptamine transporter inhibitors based on machine learning

Area of Science:

• Computational Chemistry
• Pharmacology
• Machine Learning

Background:

• Selective serotonin reuptake inhibitors (SSRIs) are crucial for treating depression.
• Ligand-based activity prediction using machine learning is an emerging area in drug discovery.

Purpose of the Study:

• To develop and evaluate machine learning models for predicting serotonin transporter (SERT) inhibitors.
• To identify key molecular features associated with SERT inhibitory activity.

Main Methods:

• Screening and preprocessing SERT inhibitor data from the ChEMBL database.
• Building 16 prediction models using four machine learning algorithms (LR, SVM, RF, KNN) and four molecular fingerprints (CDK, Graph, MACCS, PubChem).
• Developing ensemble learning models based on the top-performing cross-validation models.

Main Results:

• The ensemble model VOT_CLF3 achieved high accuracy (Q=0.869) and AUC (0.919) on the test set.
• VOT_CLF3 demonstrated strong performance on both balanced and unbalanced datasets, with high SE, SP, Q, and MCC values.
• Twelve significant molecular structural alerts for SERT inhibitors were identified (P < 0.05).

Conclusions:

• The VOT_CLF3 ensemble model is recommended for virtual screening of potential SERT inhibitors.
• The identified molecular structural alerts provide valuable insights for the rational design of novel SERT inhibitors.
• This study highlights the potential of machine learning in accelerating the discovery of antidepressant drug candidates.