CAPICE: a computational method for Consequence-Agnostic Pathogenicity Interpretation of Clinical Exome variations

Area of Science:

• Genomics
• Bioinformatics
• Computational Biology

Background:

• Exome sequencing is a standard clinical tool for genetic diagnosis.
• Identifying pathogenic Mendelian variants is often time-consuming due to limitations in current computational prediction accuracy, necessitating expert manual review.

Purpose of the Study:

• To introduce CAPICE, a novel machine-learning-based method for prioritizing pathogenic variants, including single nucleotide variants (SNVs) and short insertions/deletions (InDels).
• To evaluate CAPICE's performance against existing state-of-the-art prediction methods.

Main Methods:

• Development of a machine-learning model named CAPICE.
• Benchmarking CAPICE against general (CADD, GAVIN) and consequence-type-specific (REVEL, ClinPred) prediction tools using rare and ultra-rare variants.

Main Results:

• CAPICE demonstrates superior performance in prioritizing pathogenic variants compared to existing leading computational methods.
• The method shows high accuracy for both rare and ultra-rare genetic variants.

Conclusions:

• CAPICE offers a significant advancement in the accuracy and efficiency of identifying pathogenic variants from exome sequencing data.
• The tool is readily integrable into clinical diagnostic workflows via score files, command-line software, or a web service with API access.