VETA

VETA (Variant Effect Tool Assessment) benchmarks 39 computational variant pathogenicity prediction algorithms using annotated Variant Call Format (VCF) files to improve classification of missense variants associated with hypertrophic cardiomyopathy (HCM).

Key Features:

• Comprehensive Benchmarking: Systematically assesses and compares the performance of 39 computational prediction tools on variant pathogenicity.
• VCF-based Analysis: Analyzes annotated Variant Call Format (VCF) files to evaluate tool predictions against variant annotations.
• Evaluation Dataset: Uses high-confidence HCM-causing missense variants and benign variants as the discrimination benchmark.
• HCM-Specific Calibration: Calibrates prediction-score thresholds specifically for hypertrophic cardiomyopathy (HCM) to improve discrimination between disease-causing and benign variants.
• Bias Mitigation: Addresses potential biases including circularity type I during benchmarking.
• Identification of Top Performers: Identifies ClinPred, MISTIC, FATHMM, MPC, and MetaLR as top-performing tools for discriminating HCM-associated missense variants.
• Variant Prioritization: Combines insights from top-performing tools to prioritize unknown missense variants for closer clinical follow-up.

Scientific Applications:

• Algorithm Benchmarking for HCM: Enables comparative evaluation of pathogenicity prediction algorithms in the context of hypertrophic cardiomyopathy.
• Threshold Calibration: Provides HCM-specific score thresholds to improve variant classification accuracy.
• VUS and Missense Prioritization: Prioritizes variants-of-unknown-significance (VUS) and other unknown missense variants for targeted clinical investigation.
• Clinical Decision Support: Informs which variants warrant further investigation by combining predictive scores from top-performing tools.

Methodology:

Analyzes annotated VCF files and systematically compares 39 prediction tools using high-confidence HCM-causing missense and benign variants, applies HCM-specific calibration of thresholds, addresses circularity type I bias, and combines predictive scores from top-performing tools for variant prioritization.