CAGI5: Objective performance assessments of predictions based on the Evolutionary Action equation

Area of Science:

• Genomics
• Computational Biology
• Bioinformatics

Background:

• Computational methods for estimating coding variant effects often yield conflicting predictions, impacting user reliability.
• Objective performance assessments are crucial for understanding the accuracy, advantages, and limitations of these prediction tools.

Purpose of the Study:

• To evaluate the reliability of the Evolutionary Action (EA) method in predicting the fitness effects of coding mutations.
• To assess EA's performance across diverse prediction tasks within the Critical Assessment of Genome Interpretation (CAGI) framework.

Main Methods:

• Utilized the Evolutionary Action (EA) method, an untrained approach relying on homology and a formal equation to quantify variant fitness effects.
• Participated in CAGI experiments from 2011-2016 (protein activity) and CAGI5 in 2018 (clinical associations, folding stability, phenotype matching).
• Adapted the EA method to address the specific requirements of each CAGI challenge.

Main Results:

• EA submissions demonstrated consistently good performance across all evaluated CAGI challenges.
• The method showed reliable predictive capabilities for various aspects of genetic variant effects, including protein activity, clinical associations, and stability.

Conclusions:

• The Evolutionary Action (EA) method provides a reliable approach for predicting the fitness effects of genetic variants.
• EA's consistent performance across diverse challenges highlights its utility in genomic interpretation.