Gene dosage architecture across complex traits and common diseases 

Introduction : Copy Number Variants (CNVs) are genomic deletions or duplications larger than 1000 base pairs (the fundamental unit of double-stranded nucleic acids), and play a crucial role in complex traits and common diseases [1][2][3][4]. Coding CNVs offer unique molecular insights; to date, only a few studies have conducted genome-wide CNV investigations, facing challenges such as limited statistical power and focusing primarily on a subset of CNVs. In contrast, Single-Nucleotide Polymorphisms (SNPs) associated with complex traits provide indirect insight into biological functions and it is often difficult to link them to changes in gene transcription and function [5]. Knowledge Gap: CNVs architecture and its potential convergence to common-variants architecture, across traits, remains unexplored. Previous studies, including our own, have shown that measures of constraint scores (Intolerance to inactivation score or loss of function score) explain a significant proportion of the effects of coding CNVs on cognitive, and behavioral traits. It is shown that close to 50% of coding genes with low constraint score (high intolerant to inactivation), decrease cognitive ability when deleted or duplicated [6]. However, it is unclear if the biological function of genes can inform on the effects of CNVs beyond constraint metrics. Overarching Aim: Our overarching aim is to comprehensively characterize the CNV architecture of complex traits.  Hypothesis: We hypothesize that common variants and CNVs have pleiotropic effects on the same pairs of traits with similar genetic-correlations. Dataset & Analytical Design: Our outcome phenotypic measures of interest cover 28 traits, including cognition, socioeconomic status, blood assays, physical and psychological health, from 500,000 UK-Biobank participants. We use CNVs >50 kilobases, fully encompassing >1 genes. Since the population frequency (<1/10000) of most coding CNVs remains an obstacle for variant-level associations, we aggregate variants that disrupt genes involved in the same biological function for each individual. Our main steps are:      1. Defining gene-lists and aggregating CNVs within gene-lists. The gene-lists are defined                 based on gene expression specificity across tissues or cell-types and top 10% genes are considered for each gene-list. We use 3 resources to define gene-lists:                 - Whole Body Tissues: 60 tissues from FANTOM5, The Functional Annotation of Mammalian Genomes 5 project.                  - Whole Body Single Cells: 81 cell types from Human Protein Atlas (HPA).                 - Brain Single Cells: 461 brain cells from Human Brain Cell Atlas.       2. Estimating the effect sizes of CNVs on traits using burden association analysis (CNVs analysis. in aggregate), per gene-list, for deletion and duplication separately.      3.Computing burden genetic correlation between pairs of traits, using effect-sizes.      4.Comparing CNV and common variant genetic correlations Results : We observed that some traits (Townsend deprivation index, pulmonary metrics, neuroticism, cognitive ability) were associated with most tissues, while others were associated with only a few gene sets. We found that brain tissues are highly pleiotropic and have in general higher proportions of intolerant genes than non-brain tissues. Furthermore, at the cellular level, neuronal and glial cells are associated with cognitive traits the most. Precisely, among non-neuronal cells, microglia, choroid plexus, and oligodendrocytes have the strongest associations with cognitive traits. For neuronal cells, both excitatory and inhibitory neurons are linked to cognitive traits. We found that CNVs and common-variant architectures are concordant (slope(B):1.53, p=2.44e-35), based on genetic-correlations, as illustrated in the figure. Stratification based on constraint scores did not impact this concordance. 82% of traits showed negative correlations between deletion and duplication effect-size profiles.  Conclusion: We found that the majority of traits are dosage-sensitive. There is a convergent association between CNVs and common variants which proves that they have pleiotropic effects on the same pairs of traits with similar genetic-correlations. The constraint measure influences gene lists, and that impacts on pleiotropy. References: [1]: Auwerx C, Lepamets M, Sadler MC, Patxot M, Stojanov M, Baud D, Mägi R; Estonian Biobank Research Team; Porcu E, Reymond A, Kutalik Z. The individual and global impact of copy-number variants on complex human traits. Am J Hum Genet. 2022 Apr 7;109(4):647-668. doi: 10.1016/j.ajhg.2022.02.010. Epub 2022 Mar 2. PMID: 35240056; PMCID: PMC9069145. [2]: Auwerx C, Jõeloo M, Sadler MC, Tesio N, Ojavee S, Clark CJ, Mägi R; Estonian Biobank Research Team; Reymond A, Kutalik Z. Rare copy-number variants as modulators of common disease susceptibility. Genome Med. 2024 Jan 8;16(1):5. doi: 10.1186/s13073-023-01265-5. PMID: 38185688; PMCID: PMC10773105. [3]: Kendall KM, Bracher-Smith M, Fitzpatrick H, Lynham A, Rees E, Escott-Price V, Owen MJ, O'Donovan MC, Walters JTR, Kirov G. Cognitive performance and functional outcomes of carriers of pathogenic copy number variants: analysis of the UK Biobank. Br J Psychiatry. 2 May;214(5):297-304. doi: 10.1192/bjp.2018.301. Epub 2 Feb 15. PMID: 30767844; PMCID: PMC6520248. [4]: Hujoel, Margaux LA, et al. "Influences of rare copy-number variation on human complex traits." Cell 185.22 (2022): 4233-4248. [5]: Tam, V., Patel, N., Turcotte, M. et al. Benefits and limitations of genome-wide association studies. Nat Rev Genet 20, 467–484 (2). https://doi.org/10.1038/s41576--0127-1 [6]: Huguet, G.et al.Genome-wide analysis of gene dosage in 24,092 individuals estimates that 10,000genes modulate cognitive ability.Mol Psychiatry(2021).