Classification models for KCNQ1 variants distinguish functional and trafficking effects to enhance pathogenicity...

Area of Science:

• Genetics
• Molecular Biology
• Computational Biology

Background:

• Congenital long QT syndrome (LQTS) is a common genetic arrhythmia.
• Missense variants in the KCNQ1 potassium channel are the primary cause of LQTS.
• Variant interpretation is crucial for understanding disease mechanisms and patient risk.

Purpose of the Study:

• To develop predictive models for KCNQ1 variant effects.
• To integrate machine learning predictions with biophysical data for enhanced accuracy.
• To interpret variants of uncertain significance and AlphaMissense-ambiguous variants.

Main Methods:

• Developed random forest classifiers to predict KCNQ1 electrophysiology and trafficking metrics.
• Integrated large machine learning model predictions with protein-specific biophysical values.
• Applied classifiers to ClinVar and AlphaMissense datasets to generate global dysfunction and mistrafficking scores.

Main Results:

• Classifiers integrating both feature sets outperformed those using single sets.
• Global scores effectively distinguished benign from pathogenic KCNQ1 variants.
• Scores complemented AlphaMissense predictions, linking variants to LQTS mechanisms.

Conclusions:

• The developed approach accurately predicts KCNQ1 variant effects on protein function and trafficking.
• This method aids in the interpretation of genetic variants associated with LQTS.
• The generalizable approach can be applied to other ion channels, emphasizing the need for systematic benchmarking.