Modeling hERG Channel Liability: From Structural Insight to Highly Accurate Qualitative and Quantitative Models.

Area of Science:

• Pharmacology
• Computational Chemistry
• Drug Discovery

Background:

• Drug-induced QT interval prolongation, often due to hERG channel blockade, is a major safety concern leading to drug recalls.
• Early detection of potential hERG channel blockers is vital to reduce late-stage drug development failures.

Purpose of the Study:

• To understand ligand-channel interactions within the hERG channel.
• To develop predictive computational models for identifying hERG channel blockers early in drug discovery.

Main Methods:

• Analysis of cryo-electron microscopy (cryo-EM) structures and pharmacophore models of hERG channels.
• Development of quantitative classification and regression models using customized atom-type descriptors.
• Training and validation of models on large, curated datasets of chemical compounds.

Main Results:

• Regression model achieved AAE of 0.383 and RMSEP of 0.548 on test sets.
• Classification model achieved an ROC AUC of 0.88.
• External validation yielded an AAE of 0.50 log units for the regression model.

Conclusions:

• Developed predictive models significantly aid in early assessment of cardiotoxicity risks associated with hERG channel blockade.
• These computational strategies support safer and more efficient drug development by identifying liabilities early.
• Understanding hERG ligand-channel interactions provides mechanistic insights into drug interactions.