Polygenic Prediction of Depression Liability Reveals Broad Genetic Associations Across Development, Families, and Populations
- 1Herasight Research
Abstract
Polygenic prediction of depression is limited by the depth of depression phenotyping available at biobank scale as well as sample size. Here we develop a depression score combining information from depression and genetically correlated psychiatric traits using phenotype reconstruction, multi-trait genetic integration and Bayesian effect estimation. Compared with the strongest publicly available depression-only score (Psychiatric Genomics Consortium 2025, excluding 23andMe), the liability-R2 for depression in UK Biobank rose by 49.7% and within-family standardized associations with latent internalizing were up to 54% larger. Across 2,328 childhood and adolescent outcomes in the ABCD study, the association profile of the integrated score was almost identical to that of the depression-only score (r = 0.99), while being uniformly stronger. 1,494 outcomes related to mood, behavior, social functioning, hobbies and sleep were associated with the score -- 423 of them within-family. In ABCD, adolescents in the highest score decile were 3.5 times as likely as those in the lowest decile to report suicidal thoughts by age 14 (23% versus 7%) and 1.9 times as likely to report nicotine or cannabis use by age 17 (47% versus 25%). In adults, the score predicted worse general health, chronic pain, and life dissatisfaction. Associations with psychiatric symptoms were substantially stronger in girls than in boys (boys' coefficients 0.68 times girls'), particularly for girl-typical symptoms, and showed no moderation by household resources, hardship or deprivation that remained significant after correction for multiple testing. Prediction was attenuated in line with genetic distance from training data for Hispanic and African American children. Depression prediction can thus be strengthened substantially by integrating the internalizing spectrum, without detectable loss of specificity or within-family predictive ability.
Related
- The landscape of polygenic score prediction accuracy and transportability across the phenome in ancestrally diverse populations
- Novel type 1 diabetes polygenic scores improve identification of type 1 diabetes in diverse populations
- Lysosomal polygenic risk score in Parkinson's disease across populations.
- Hybrid risk scores integrating polygenic and clinical variables for endometriosis prediction
- Polygenic liability for major depression, schizophrenia and bipolar disorder and treatment outcome across major psychiatric disorders
BibTeX
@misc{wolfram2026polygenic,
title = {{Polygenic Prediction of Depression Liability Reveals Broad Genetic Associations Across Development, Families, and Populations}},
author = {T. Wolfram and S. M. Moore and J. Anomaly and J. Schleede and M. Ahangari},
howpublished = {medRxiv},
year = {2026},
month = nov,
doi = {10.64898/2026.11.06.26367433},
url = {https://www.medrxiv.org/content/10.64898/2026.11.06.26367433v1},
note = {CC BY-NC-ND},
}
RIS
TY - UNPB
TI - Polygenic Prediction of Depression Liability Reveals Broad Genetic Associations Across Development, Families, and Populations
AU - T. Wolfram
AU - S. M. Moore
AU - J. Anomaly
AU - J. Schleede
AU - M. Ahangari
PY - 2026
DA - 2026/11/06
PB - medRxiv
DO - 10.64898/2026.11.06.26367433
UR - https://www.medrxiv.org/content/10.64898/2026.11.06.26367433v1
AB - Polygenic prediction of depression is limited by the depth of depression phenotyping available at biobank scale as well as sample size. Here we develop a depression score combining information from depression and genetically correlated psychiatric traits using phenotype reconstruction, multi-trait genetic integration and Bayesian effect estimation. Compared with the strongest publicly available depression-only score (Psychiatric Genomics Consortium 2025, excluding 23andMe), the liability-R2 for depression in UK Biobank rose by 49.7% and within-family standardized associations with latent internalizing were up to 54% larger. Across 2,328 childhood and adolescent outcomes in the ABCD study, the association profile of the integrated score was almost identical to that of the depression-only score (r = 0.99), while being uniformly stronger. 1,494 outcomes related to mood, behavior, social functioning, hobbies and sleep were associated with the score -- 423 of them within-family. In ABCD, adolescents in the highest score decile were 3.5 times as likely as those in the lowest decile to report suicidal thoughts by age 14 (23% versus 7%) and 1.9 times as likely to report nicotine or cannabis use by age 17 (47% versus 25%). In adults, the score predicted worse general health, chronic pain, and life dissatisfaction. Associations with psychiatric symptoms were substantially stronger in girls than in boys (boys' coefficients 0.68 times girls'), particularly for girl-typical symptoms, and showed no moderation by household resources, hardship or deprivation that remained significant after correction for multiple testing. Prediction was attenuated in line with genetic distance from training data for Hispanic and African American children. Depression prediction can thus be strengthened substantially by integrating the internalizing spectrum, without detectable loss of specificity or within-family predictive ability.
ER -