Deep learning-based discovery of compounds for blood pressure lowering effects

Area of Science:

• Pharmacology
• Computational Chemistry
• Drug Discovery

Background:

• Hypotensive side effects from medications pose a significant clinical challenge.
• Deep learning offers a promising approach to predict compound biological activity and identify potential drug side effects by analyzing complex datasets.

Purpose of the Study:

• To develop and validate a deep learning model for predicting blood pressure-lowering effects of chemical compounds.
• To identify novel antihypertensive agents and mitigate risks of drug-induced hypotension.

Main Methods:

• Established a deep learning model trained on compounds affecting blood pressure.
• Predicted blood pressure-lowering effects for 26,000 compounds.
• Validated predictions through literature comparison, molecular docking, and in vivo studies.

Main Results:

• The model accurately predicted antihypertensive properties for 30 out of 50 selected molecules, including known drugs.
• Molecular docking studies indirectly confirmed the model's predictive accuracy.
• Salaprinol was identified as a novel ACE1 inhibitor with significant blood pressure-lowering effects in canines.

Conclusions:

• A highly accurate deep learning model for predicting antihypertensive activity was developed.
• The model facilitates the discovery of new blood pressure-lowering drugs and helps prevent adverse hypotensive effects in patients.
• This approach supports safer clinical medication and accelerates antihypertensive drug development.