CardioVar: a machine learning framework for pathogenicity prediction of cardiomyopathy genetic variants

Area of Science:

• Genetics
• Cardiology
• Bioinformatics

Background:

• Cardiomyopathy-associated genetic variants require accurate pathogenicity prediction for clinical interpretation.
• Existing pan-disease models may not fully capture the nuances of cardiomyopathy-specific variants.

Purpose of the Study:

• To introduce CardioVar, a machine learning framework for predicting the pathogenicity of genetic variants associated with cardiomyopathy.
• To provide a rapid, disease-focused classification score to aid variant interpretation in cardiogenetic workflows.

Main Methods:

• Development of a machine learning framework (CardioVar).
• Application of the framework to predict pathogenicity of cardiomyopathy-associated genetic variants.
• Evaluation of CardioVar's performance in prospective case study settings.

Main Results:

• CardioVar provides a rapid and disease-focused classification score for genetic variants.
• The framework supports variant interpretation within cardiac-related genes.
• CardioVar demonstrated an 85% concordance with conventional tertiary analysis results in prospective case studies.

Conclusions:

• CardioVar offers a valuable tool for supporting variant interpretation in cardiogenetic workflows.
• This cardiomyopathy-specific approach complements existing pan-disease models.
• The framework shows promising concordance with established analysis methods.