Using DNA to predict behaviour problems from preschool to adulthood.

Area of Science:

• Behavioral genetics
• Psychiatric genetics
• Developmental psychology

Background:

• The DNA revolution aims to predict and prevent health issues.
• Genome-wide polygenic scores (GPS) are being explored for predicting behavioral problems.
• Improving prediction accuracy is crucial for early intervention.

Purpose of the Study:

• To enhance the prediction of behavioral problems using genome-wide polygenic scores (GPS).
• To investigate the impact of cross-age and cross-rater composites on prediction accuracy.
• To explore a multi-GPS approach incorporating various behavioral and psychiatric traits.

Main Methods:

• Utilized data from 3,065 genotyped individuals in the Twins Early Development Study.
• Assessed hyperactivity, conduct, emotional, and peer problems longitudinally (ages 2-21) via parent, teacher, and self-reports.
• Employed penalized regression models with 15 genome-wide association studies-derived GPS to predict behavior problem composites, creating multi-trait GPS.

Main Results:

• Multi-GPS prediction of behavioral problems increased from <2% to up to 6% variance for cross-age and cross-rater composites.
• Twin heritability estimates showed a similar pattern, increasing from <40% to 83%.
• The multi-GPS approach demonstrated improved predictive power compared to single-GPS models.

Conclusions:

• Combining multiple GPS, cross-age, and cross-rater composites significantly improves the prediction of behavioral problems.
• While effect sizes are modest (up to 6%), the approach offers enhanced predictive utility.
• This methodology can be applied to any genotyped sample for developing more accurate multi-trait GPS predictors of behavioral issues.