MARGINAL: An Automatic Classification of Variants in BRCA1 and BRCA2 Genes Using a Machine Learning Model

Area of Science:

• Genetics
• Bioinformatics
• Machine Learning

Background:

• Next-generation sequencing (NGS) generates numerous genetic variants daily, complicating interpretation for hereditary diseases.
• Inadequate interpretation of rare BRCA1 and BRCA2 germline variants hinders clinical treatment decisions.

Purpose of the Study:

• To introduce MARGINAL 1.0.0, a machine learning (ML)-based software for interpreting rare BRCA1 and BRCA2 germline variants.
• To automate variant classification according to American College of Medical Genetics and Genomics and Association for Molecular Pathology (ACMG-AMP) criteria.

Main Methods:

• Annotation of BRCA1 and BRCA2 variants from diverse sources.
• Automated implementation of ACMG-AMP criteria for variant classification.
• Development and comparison of eight ML algorithms in a two-classifier system to maximize accuracy.

Main Results:

• The MARGINAL software achieved high predictive performance with maximum accuracies of 92% and 98% for its two classifiers.
• Recall reached 92% and 98%, while specificity was 90% and 98% for the respective classifiers.
• The ML model demonstrated robust capabilities in classifying rare germline variants.

Conclusions:

• MARGINAL 1.0.0 offers an automated, gene and disease-specific solution for clinical variant evaluation.
• This ML-based approach can minimize conflicting interpretations of BRCA1 and BRCA2 variants.
• The software enhances the utility of genetic information for guiding treatment decisions in hereditary diseases.