Estimating the effects of copy-number variants on intelligence using hierarchical Bayesian models

Area of Science:

• Genetics
• Neuroscience
• Bioinformatics

Background:

• Copy-number variations (CNVs) are structural genome changes with challenging-to-estimate phenotypic impacts.
• Previous methods aggregated CNV effects using mutation intolerance scores (pLI), indirectly estimating individual CNV impacts.
• Rare and unique CNVs pose challenges for direct analysis of their effects on traits like intelligence.

Purpose of the Study:

• To develop a method for directly estimating the individual effects of rare CNVs on intelligence.
• To utilize genomic annotation data to inform a Bayesian modeling approach for CNV analysis.
• To improve the prediction accuracy of CNV consequences on intellectual quotient (IQ).

Main Methods:

• Developed a hierarchical Bayesian model named HBIQ to directly estimate individual rare CNV effects on intelligence.
• Integrated genomic annotation data, including mutation impact information, to define prior probabilities for the model.
• Applied HBIQ to analyze CNV deletions and duplications across three independent datasets.

Main Results:

• Identified specific genomic regions with CNVs that have significant deleterious effects on IQ, validating some known associations.
• Demonstrated that HBIQ can identify deleterious CNVs even those with a zero pLI score, and vice versa.
• Achieved higher out-of-sample concordance (78%) in predicting consequences of known recurrent CNVs compared to previous aggregation methods.

Conclusions:

• Hierarchical Bayesian models offer a direct approach to estimating individual rare CNV effects on intelligence.
• HBIQ provides a more accurate and nuanced understanding of CNV impacts on IQ than aggregation-based methods.
• This approach enhances the ability to predict the phenotypic consequences of structural genomic variations.