Determinants of pleiotropy and monotonic gene dosage responses across human traits

Sayeh Kazem, Kuldeep Kumar, Martineau Jean‐Louis, Guillaume Huguet, Thomas Renne, Zohra Saci, Worrawat Engchuan, Omar Shanta, Bhooma Thiruvahindrapuram, Jeffrey R. MacDonald, Celia M.t. Greenwood, Stephen W. Scherer, Laura Almasy, Jonathan Sebat, David C. Glahn, Guillaume Dumas, Sébastien JacquemontmedRxiv 2025 · DOI

Abstract

Copy Number Variants (CNVs), involving the deletion or duplication of genomic segments, significantly influence human phenotypic variation. Using a novel framework termed Functional Burden analysis (FunBurd), we investigated the effects of rare CNVs across 43 human traits in the UK Biobank. Our analysis revealed that brain-related gene sets exhibit the highest levels of pleiotropy, an effect largely independent of general genetic constraint. We found that deletions and duplications frequently affect distinct biological pathways, with true "monotonic" responses—where deletions and duplications have opposite phenotypic effects—occurring in only 4% of cases. These findings highlight the complex, non-linear relationship between gene dosage and human health.

Introduction

While single-nucleotide variants (SNVs) are frequently the focus of genome-wide association studies, Copy Number Variants (CNVs) represent a potent source of genetic variation by altering gene dosage. Rare CNVs are well-documented drivers of neurodevelopmental and psychiatric disorders, yet their broader impact on complex human traits remains less understood. A central question in dosage genetics is monotonicity: whether the phenotypic response to a deletion is mirrored by an opposite response to a duplication. This study introduces a systematic approach to map these dosage effects across the genome and a wide array of physical and clinical traits.

Methodology

To overcome the statistical challenges of studying rare variants, we developed Functional Burden analysis (FunBurd). This framework aggregates rare CNVs into 172 functional gene sets based on tissue-specific and cell-type-specific expression patterns.

Key Components:

  • Top Decile Expression Proportion (TDEP): A metric used to define gene sets by identifying genes with specialized expression in specific tissues or cell types, utilizing data from FANTOM, GTEx, and the Human Protein Atlas.
  • P-Jaccard: A permutation-based method designed to account for the high degree of overlap between related gene sets, ensuring that associations are not artifactually inflated.
  • Normative Constraint Modeling: We integrated LOEUF scores from gnomAD to distinguish between effects driven by a gene's specific biological function versus its general intolerance to mutation (genetic constraint).

Results and Discussion

Our analysis identified significant associations between CNVs and all 43 traits studied. Notably, gene sets active in the brain and neuronal cell types demonstrated the highest levels of pleiotropy, affecting diverse traits ranging from cognitive measures to physical markers like height and blood composition. While genetic constraint explains much of the functional enrichment seen in other tissues, brain-related processes remained significantly associated with traits even after adjusting for constraint, suggesting a unique sensitivity of the human phenome to brain gene dosage.

Asymmetry in Dosage Responses

A key finding of this study is the lack of symmetry between deletions and duplications. Deletions proved more potent, showing significant associations in 22% of tests compared to 13.6% for duplications. Furthermore, the "mirror effect" of monotonicity was remarkably rare; only 4% of gene-dosage responses were significant and opposite in direction. Instead, 84% of traits showed a negative correlation between deletion and duplication effect sizes, reinforcing the idea that biological systems are usually sensitive to one direction of dosage change but not the other.

Functional Convergence and Divergence

We observed limited functional overlap between the effects of CNVs and common variants. Even within CNVs, deletions and duplications often impacted different biological pathways. For instance, in the cortex, deletions in sensorimotor regions were primarily linked to cognitive performance, whereas duplications in higher-order association regions showed stronger influences.

Conclusion

This study provides a comprehensive map of how gene dosage shapes human traits. By separating the contributions of genetic constraint from specific gene functions, we show that brain-related genes are primary drivers of phenotypic complexity. The widespread lack of monotonicity suggests that the biological consequences of gene deletions and duplications are fundamentally distinct, challenging simple "volume knob" models of gene expression and highlighting the need for nuanced interpretations of rare variant data in clinical genetics.

Limitations

The use of the UK Biobank introduces a "healthy volunteer" bias, potentially underestimating the impact of the most severe CNVs. Additionally, the cohort's predominantly European ancestry necessitates further research in more diverse populations to confirm the universality of these dosage rules.

Summary updated

Abstract

All 43 traits showed FDR significant associations with CNVs. Brain tissue and neuronal cell-types showed the highest levels of pleiotropy. Most of the functional gene set associations could, in part, be explained by genetic constraint, except for brain related processes. Shared genetic contributions between pairs of traits were concordant across types of variants, but on average 2-fold higher, for rare CNVs and SNVs compared to common variants.Functional enrichment across traits found limited overlap between CNVs and common variants. Moreover, the effects of deletions and duplications were negatively correlated for most traits.In conclusion, we present new methods to separate the contributions of genetic constraint and gene function to the associations of CNVs with complex traits. Overall, the functional convergence between different types of variants -even between deletions and duplications-remains limited.
BibTeX
@misc{kazem2025determinants,
  author = {Sayeh Kazem and Kuldeep Kumar and Martineau Jean‐Louis and Guillaume Huguet and Thomas Renne and Zohra Saci and Worrawat Engchuan and Omar Shanta and Bhooma Thiruvahindrapuram and Jeffrey R. MacDonald and Celia M.T. Greenwood and Stephen W. Scherer and Laura Almasy and Jonathan Sebat and David C. Glahn and Guillaume Dumas and Sébastien Jacquemont and Sébastien Jacquemont},
  title = {{Determinants of pleiotropy and monotonic gene dosage responses across human traits}},
  year = {2025},
  doi = {10.1101/2025.02.25.25322833},
  howpublished = {medRxiv},
}
Figure