A multilayer genomic framework linking insomnia to glymphatic system function through pleiotropic mechanisms at 17q21.31

A multilayer genomic framework linking insomnia to glymphatic system function through pleiotropic mechanisms at 17q21.31 | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Article A multilayer genomic framework linking insomnia to glymphatic system function through pleiotropic mechanisms at 17q21.31 Chenxu Xiao, Jing Shen, Yuxuan Shen This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-8585990/v1 This work is licensed under a CC BY 4.0 License Status: Under Review Version 1 posted You are reading this latest preprint version Abstract Sleep plays a critical role in brain waste clearance, yet whether insomnia shares a genetic basis with the glymphatic system (GS)—a glia-dependent perivascular pathway involved in metabolite removal—remains unclear. Here, we integrated large-scale genome-wide association studies (GWAS) of insomnia with diffusion tensor imaging along the perivascular space (DTI–ALPS), an imaging-derived proxy of GS function, across two independent cohorts. Polygenic enrichment analyses revealed localized genetic sharing between insomnia and multiple ALPS phenotypes despite minimal genome-wide genetic correlation. Conjunctional false discovery rate and Bayesian colocalization analyses identified shared causal signals at 17q21.31, a pleiotropic locus encompassing the MAPT inversion region. Transcriptome-wide association, gene-level fine-mapping, and summary-data Mendelian randomization converged on HEXIM1 , ACBD4 , EFTUD2 , and MAPT as shared genes influencing both insomnia and GS function. Functional characterization showed that these genes were enriched across multiple brain regions and cell types, including neurons, astrocytes, microglia, oligodendroglia, and vascular-associated cells. Notably, gene-level effects exhibited regional and phenotype-specific heterogeneity. Together, our findings demonstrate that insomnia and glymphatic function converge through a context-dependent genetic architecture centered on 17q21.31, implicating neuroglial pathways relevant to protein clearance and Alzheimer’s disease vulnerability. Biological sciences/Genetics/Genetic association study/Genome-wide association studies Health sciences/Neurology/Neurological disorders/Sleep disorders Insomnia Glymphatic system DTI-ALPS Pleiotropy 17q21.31 locus Integrative genomics Figures Figure 1 Figure 2 Figure 3 Figure 4 Figure 5 INTRODUCTION Sleep is fundamental to brain homeostasis, metabolic maintenance, and cognitive integrity( 1 – 4 ). Insomnia, one of the most prevalent sleep disorders, affects nearly one-third of adults( 5 ) and has been shown by large-scale genome-wide association studies (GWAS) to involve a highly polygenic architecture spanning diverse biological pathways( 6 – 9 ). Converging experimental and imaging evidence demonstrates that sleep promotes cerebrospinal fluid (CSF)–interstitial fluid exchange and facilitates the clearance of neurotoxic metabolites, whereas sleep loss diminishes CSF influx and impairs the removal of proteins such as amyloid-β (Aβ) and phosphorylated tau (pTau)( 10 , 11 ). These observations strongly suggest a mechanistic interface between sleep regulation and brain waste clearance. The glymphatic system (GS) constitutes a perivascular, glia-dependent transport network that drives the clearance of metabolic by-products, including Aβ and pTau( 12 , 13 ). Despite strong physiological evidence for sleep–GS coupling, it remains unknown whether insomnia and GS function share underlying genetic determinants, and whether any such overlap exhibits anatomical or molecular specificity. One major reason is that large-scale GWAS capable of resolving the genetic architecture of GS function have long been unavailable( 14 , 15 ). This barrier was overcome only recently with two population-level GWAS of MRI-derived diffusion tensor imaging along-the-perivascular-space (DTI-ALPS) phenotypes—one characterizing hemispheric ALPS indices in more than 31,000 individuals( 14 ), and another defining anatomically specific ALPS measures in more than 40,000 individuals( 15 ). ALPS serves as a noninvasive imaging proxy of GS function by quantifying water diffusivity along perivascular white-matter tracts, thereby reflecting interstitial fluid transport mediated by astrocytic endfeet and perivascular pathways—the core anatomical substrates of the GS function( 14 , 15 ). These datasets therefore provide, for the first time, genetically informative and scalable proxies of GS function, enabling systematic evaluation of whether insomnia and GS converge at the genomic level and whether such convergence is spatially or biologically selective. In addition, prior studies suggest that GS activity primarily involves astrocytes and other neurovascular cell populations( 12 , 16 ), raising the question of whether insomnia–GS shared genetic signals preferentially map onto specific brain tissues or neural cell types. Finally, because GS function plays a central role in clearing Aβ and pTau( 12 , 13 ), it is critical to determine whether genetic convergence between insomnia and GS aligns with regions or pathways vulnerable to neurodegenerative protein aggregation. METHODS Data Sources Insomnia GWAS Dataset The summary statistics for insomnia were obtained from a large-scale GWAS conducted by Lane et al. in the UK Biobank, with external validation in independent cohorts( 17 ). The primary analysis was based on 453,379 individuals of European ancestry who self-reported their sleep patterns. For the GWAS, insomnia symptoms were defined using the question “Do you have trouble falling asleep at night, or do you wake up in the middle of the night?”. Participants were categorized into two primary phenotypes: those reporting “usually” (frequent insomnia) and those reporting “sometimes” or “usually” (any insomnia). This genetic association was robustly replicated in independent samples, including self-reported insomnia cases from the HUNT study and physician-diagnosed insomnia cases from the Partners Biobank, as well as in objective measures of sleep efficiency and duration from accelerometer data within the UK Biobank. GS Function GWAS Datasets We leveraged two recently published, comprehensive GWAS datasets on GS function as measured by the DTI-ALPS index. For the analysis of hemispheric ALPS indices, we used summary statistics from Huang et al.( 14 ). This study analyzed data from 31,021 participants of white British ancestry in the UK Biobank, calculating separate indices for the left and right cerebral hemispheres (Left_ALPS, Right_ALPS) and their average (Mean_ALPS). For the analysis of regional ALPS indices, we utilized data from Ran et al.( 15 ). This research employed DTI-ALPS data from 40,486 European-ancestry individuals in the UK Biobank to define four distinct phenotypes based on anatomical location: anterior (aALPS), middle (mALPS), posterior (pALPS), and total (tALPS) ALPS indices. Quality Control Prior to our analysis, the publicly available summary statistics underwent stringent quality control procedures. For the UK Biobank data, this included excluding individuals of non-European ancestry and those with incomplete genotype or phenotype data. At the variant level, single nucleotide polymorphisms (SNPs) were filtered based on minor allele frequency (MAF < 0.01), Hardy-Weinberg equilibrium ( P < 1 × 10 − 6 ), and imputation quality score (INFO < 0.8). All genomic coordinates were harmonized to the GRCh37/hg19 reference assembly. The resulting dataset showed no significant inflation due to population stratification (LDSC intercept ≈ 1.005), confirming its suitability for downstream integrative analyses. The full inclusion and exclusion criteria are provided in Supplementary Table 1, and an overview of the study design and analytic workflow is presented in Fig. 1 . Insert Fig. 1 Spatial Transcriptomics (ST) Data for gsMap Analysis To spatially contextualize the shared genetic signatures between insomnia and GS function, and to evaluate their relevance to the neuropathological continuum, we employed two complementary ST datasets. Human AD Pathology Dataset for Pathological Contextualization To test the hypothesis that insomnia may influence AD risk through impairing GS-mediated clearance of pathological proteins, we leveraged ST data from postmortem AD brains( 18 ). This dataset encompasses tissue sections from the inferior temporal cortex (ITC) of donors with advanced AD pathology (Braak V-VI) and normal controls, profiled using the 10x Genomics Visium platform. The key rationale for using this AD dataset is that it allows us to map the insomnia-GS shared genes directly onto the spatial epicenters of Aβ and pTau pathology. A significant enrichment in these pathological microenvironments would provide compelling spatial evidence linking our discovered genetic overlap to the core proteinopathic processes of AD. 2. Mouse Embryonic Development Dataset for Investigating Developmental Origins To explore whether the insomnia-GS genetic overlap is embedded within fundamental neurodevelopmental programs, we analyzed data from the Mouse Organogenesis Spatiotemporal Transcriptomic Atlas (MOSTA)( 19 ). We specifically focused on the embryonic day 16.5 (E16.5) sample (E16.5_E1S1.MOSTA.h5ad). The E16.5 stage represents a peak period of murine neurogenesis, gliogenesis, and overall brain architecture establishment. By examining the enrichment of insomnia-GS genes at this developmental critical window, we aimed to determine if their shared biological influence originates during early brain construction, potentially setting the stage for lifelong brain function and vulnerability to sleep and clearance disorders. Bivariate causal mixture model (MiXeR) analysis We used MiXeR (v2.2.1) to quantify the polygenic architecture of insomnia and each ALPS phenotype and to estimate the extent of their shared causal variant sets( 20 – 22 ). MiXeR applies a Gaussian mixture model to GWAS summary statistics to infer key parameters of genetic architecture. In MiXeR, π denotes the proportion of SNPs with non-zero causal effects (polygenicity), and σ²β represents the variance of causal SNP effect sizes. For bivariate modeling, MiXeR additionally estimates π₁₂, the proportion of shared causal SNPs between two traits, and ρ_beta, the correlation of effect sizes among these shared causal variants. From these quantities, MiXeR also derives a model-based estimate of the genetic correlation (rg_MiXeR), which reflects the genetic correlation attributable specifically to shared causal components. Summary statistics for insomnia and all seven ALPS phenotypes were subjected to unified quality control, and linkage disequilibrium (LD) was estimated using the 1000 Genomes Phase 3 European reference panel. Before model fitting, summary statistics were divided by chromosome using the split_sumstats function. For each trait, we first performed univariate model fitting using mixer.py fit1 to obtain trait-specific architecture parameters (π and σ²β), which were subsequently used as priors in bivariate modeling with mixer.py fit2 to estimate shared versus trait-specific causal components (π₁₂ and ρ_beta). Model fit was evaluated relative to nested minimal-overlap (genetic-correlation–only) and full-overlap models using AIC and BIC, ensuring that polygenic sharing was statistically justified rather than a modeling artifact. To enhance robustness, all analyses were repeated across 20 randomly sampled SNP subsets from the LD reference panel, and reported estimates represent averaged results. Final summaries, including Venn diagrams and Dice coefficients quantifying the degree of shared causal variants, were generated using the mixer_figures.py module. Genetic analysis integrating pleiotropy and functional annotation (GPA) We applied GPA( 23 ) to enhance the detection of shared risk variants across insomnia and each ALPS phenotype. GPA is a probabilistic mixture-model framework that jointly analyzes GWAS summary statistics from multiple related traits and integrates functional annotations to improve the prioritization of associated SNPs. GPA models the distribution of GWAS p-values using a latent categorical mixture. For a single trait, p-values are assumed to arise from a mixture of a uniform distribution U(0, 1) for null SNPs and a Beta (α, 1) distribution (0 < α < 1) for non-null SNPs, reflecting the tendency of associated variants to generate small p-values. For two-trait analyses (K = 2), SNPs are assigned to four latent association classes: non-associated with either trait (00), associated only with trait 1 ( 10 ), associated only with trait 2 (01), or associated with both traits ( 11 ). Model parameters were estimated via maximum likelihood. Following model fitting, we computed local false discovery rates (FDR) for each SNP corresponding to its posterior probability of belonging to each association class. For each insomnia–ALPS pair, SNPs were prioritized based on local FDR thresholds to identify variants associated with insomnia, with ALPS, or jointly with both traits while controlling the global FDR. To test whether significant pleiotropy existed between trait pairs, we performed a likelihood-ratio test comparing a null model assuming independence of association states—i.e., π₁₁ = (π₁₀ + π₁₁) (π₀₁ + π₁₁) — with an alternative model allowing non-independent enrichment of jointly associated variants. The test statistic, − 2 log λ(P), asymptotically follows a χ² distribution with 1 degree of freedom. Significant deviation from the null supports true genetic pleiotropy between insomnia and the ALPS phenotype under study. Local Analysis of [co]Variant Association (LAVA) We also performed local genetic correlation analysis using LAVA( 24 ). The genome was partitioned into 2,495 approximately independent LD blocks based on the 1000 Genomes European reference panel. For each block, LAVA estimates the local genetic correlation (ρ), defined as the correlation between the genetic effect sizes of two traits within a given LD block. Local ρ values were obtained using LAVA’s bivariate model after verifying sufficient univariate signal in both traits. To account for multiple testing, we applied a Bonferroni-corrected significance threshold of P < 0.05/2495 = 2.00 × 10 − 5 . In addition, we applied the Pairwise GWAS (GWAS-PW) method to further assess whether genomic regions were shared between ALPS and insomnia( 25 ). This approach estimates the posterior probability of a locus being associated with both traits (PPA3: shared causal variant; PPA4: distinct causal variants within the same locus). We considered a region as significantly shared if the posterior probability of colocalization (PPA3) exceeded 0.8, in accordance with the original GWAS-PW recommendations. CondFDR and conjFDR To identify genetic loci jointly associated with insomnia and ALPS traits, we applied the condFDR and conjFDR framework( 26 , 27 ). This empirical Bayes approach enhances discovery power by leveraging genetic pleiotropy between traits, re-ranking SNP association p-values from a primary GWAS based on enrichment patterns observed in a secondary GWAS. We computed the condFDR for insomnia and ALPS traits, and reciprocally for ALPS traits given insomnia, using publicly available MATLAB scripts. The condFDR for a given SNP is defined as the posterior probability that the SNP is null for the primary trait, given that its p-values in both traits are as small as or smaller than the observed values. The conjFDR, defined as the maximum of the two reciprocal condFDR values, was used to identify loci significantly associated with both traits simultaneously. To minimize bias from long-range LD, we excluded the extended major histocompatibility complex (MHC) region (chromosome 6: 25–34 Mb) and the chromosomal region 8p23.1 (chromosome 8: 7.2–12.5 Mb). All test statistics were corrected for genomic inflation using intergenic SNPs, which are relatively depleted of true associations. SNPs were randomly pruned over 500 iterations using an LD threshold of r 2 > 0.1 to approximate independence and mitigate p-value inflation. A significance threshold of Benjamini–Hochberg FDR < 0.05 was applied to declare statistical significance. Independent genomic loci were defined using FUMA, based on LD blocks (r 2 < 0.6) and merged within 250 kb windows. Colocalization and fine-mapping analyses To identify shared causal variants between insomnia and ALPS phenotypes, we performed a series of Bayesian and likelihood-based colocalization and fine-mapping analyses. First, we applied the coloc R package (v5)( 28 ) to estimate the posterior probability that two traits share the same causal variant within each genomic region. Coloc evaluates five mutually exclusive models—no association with either trait (H0), association with only one trait (H1/H2), association with both traits but via distinct causal variants (H3), or association with both traits via a shared causal variant (H4). Evidence for colocalization was defined by a high posterior probability for H4 (PP.H4), typically PP.H4 > 0.5 or > 0.8 depending on stringency. Analyses were performed using GWAS summary statistics together with LD matrices derived from the 1000 Genomes European reference panel. Because single-causal-variant assumptions may be violated in regions with complex LD or multiple independent signals, we additionally used the Sum of Single Effects (SuSiE) regression framework( 29 ) implemented through coloc.susie. SuSiE decomposes trait associations into a sparse set of “single-effect” components, each corresponding to a putative causal SNP, and returns credible sets representing distinct association signals. We ran SuSiE separately for each trait using runsusie, and then used coloc.susie to assess whether credible sets from insomnia and ALPS overlapped. This approach is well suited for loci containing multiple independent causal variants. Integration with FINEMAP and echolocatoR To further fine-map causal configurations, we used FINEMAP (v1.4+)( 30 ) as implemented within the echolocatoR pipeline( 31 ). FINEMAP enumerates possible causal configurations (up to five causal variants per region by default) and computes posterior inclusion probabilities (PIP) for individual SNPs and posterior probabilities for multi-variant causal models. Colocalization was inferred when high-probability causal configurations for insomnia and ALPS shared one or more SNPs with large PIP values. All analyses used LD matrices derived from the European 1000 Genomes reference panel, and harmonized GWAS summary statistics including effect sizes, standard errors, allele frequencies, and genomic coordinates. Across methods, we prioritized colocalized signals supported by multiple approaches or yielding high-confidence metrics (e.g., PP.H4 > 0.8 for coloc, PIP > 0.95 for SuSiE or FINEMAP). Such convergent multi-method evidence was used to define the set of credible shared causal variants at each locus. Fine-mapping Of Causal eQTLs (FOCUS) To identify putative causal genes underlying the shared genetic signals between insomnia and ALPS phenotypes, we applied FOCUS( 32 ), a Bayesian framework that integrates GWAS summary statistics with tissue-specific eQTL reference panels. FOCUS models the joint distribution of GWAS and eQTL effect sizes to compute the PIP that each gene mediates the association signal within a given locus. Analyses were restricted to genomic regions showing evidence of shared genetic architecture based on conjFDR and LAVA results. For each region, we supplied FOCUS with harmonized GWAS summary statistics for insomnia and ALPS phenotypes, LD matrices derived from the 1000 Genomes Project Phase 3 European reference panel, and precomputed eQTL weight matrices from the focus.db repository, which aggregates GTEx-based expression prediction models across multiple tissues and cell types. Gene boundaries were defined using GENCODE v37 annotations( 33 , 34 ). FOCUS returns a posterior distribution over genes that are compatible with the observed association pattern, quantifying for each gene the probability of being causal. Genes with PIP > 0.8 were considered high-confidence causal candidates. These fine-mapping results were subsequently integrated with TWAS, Summary-based Mendelian Randomization (SMR), and colocalization findings to construct a convergent causal gene set for each shared insomnia–ALPS locus. Variant annotation and functional characterization To functionally characterize genetic variants associated with insomnia and ALPS phenotypes, we performed locus definition and variant-level annotation using the Functional Mapping and Annotation (FUMA) platform( 35 ). Genomic risk loci were identified based on genome-wide significant SNPs ( P < 5 × 10⁻⁸) using LD information from the 1000 Genomes Project Phase 3 European reference panel. Independent significant SNPs were defined using FUMA’s LD-clumping procedure, and signals within 250 kb of each other were merged into a single locus to capture extended association blocks. Functional annotation of all SNPs mapped to each locus was conducted through FUMA’s ANNOVAR-based framework( 36 ). We integrated multiple annotation resources to evaluate the potential functional impact of variants, including predicted deleteriousness scores and regulatory features. Variants with CADD scores > 12.37 were considered to have potentially damaging effects, while RegulomeDB scores were used to infer regulatory activity, with lower scores indicating stronger evidence for regulatory function. Additional annotations, including chromatin states, eQTL mappings, and positional mappings to nearby genes, were incorporated to prioritize variants with plausible biological relevance to insomnia–ALPS shared genetic architecture. TWAS analyses using S-PrediXcan and S-MultiXcan We performed TWAS to identify genes whose genetically predicted expression is associated with insomnia and each ALPS phenotype, using S-PrediXcan and S-MultiXcan( 37 , 38 ). S-PrediXcan was applied to estimate gene–trait associations in whole blood using GTEx v8 pre-trained expression prediction models. This approach integrates GWAS summary statistics with cis-eQTL weights to infer whether genetically regulated gene expression contributes to trait variation. To improve power by leveraging shared regulatory effects across brain regions, we further applied S-MultiXcan, which jointly analyzes expression prediction models from 13 GTEx v8 brain tissues. These tissues included: Brain_Putamen_basal_ganglia, Brain_Frontal_Cortex_BA9, Brain_Hippocampus, Brain_Cerebellum, Brain_Anterior_cingulate_cortex_BA24, Brain_Caudate_basal_ganglia, Brain_Cerebellar_Hemisphere, Brain_Nucleus_accumbens_basal_ganglia, Brain_Spinal_cord_cervical_c-1, Brain_Amygdala, Brain_Substantia_nigra, Brain_Cortex, and Brain_Hypothalamus. Together, these GTEx v8 models encompass 838 donors and 15,201 RNA-seq samples( 39 ). All analyses used LD covariance matrices matched to the GTEx prediction models, derived from the 1000 Genomes Phase 3 European reference panel, and were conducted using the default settings of the PrediXcan/S-MultiXcan framework. Multiple testing correction was performed using the Benjamini–Hochberg FDR, and genes with FDR-significant associations were retained as TWAS-prioritized candidates. SMR and Heterogeneity in Dependent Instruments (HEIDI) testing To evaluate whether genetically regulated gene expression exerts causal effects on insomnia and ALPS phenotypes, we applied the SMR framework( 40 ). SMR integrates GWAS and cis-eQTL summary statistics to estimate the effect of gene expression on complex traits under an instrumental-variable (IV) paradigm, using the top cis-eQTL of each gene as the IV. This approach enables causal inference without requiring individual-level genotype or transcriptomic data. To distinguish true causal relationships from spurious associations driven by LD, we performed the HEIDI test as implemented in the SMR software. Associations showing no evidence of heterogeneity (p_HEIDI > 0.01) were interpreted as consistent with a shared causal variant, whereas signals failing the HEIDI test were excluded as potential LD-induced artifacts. We conducted SMR analyses using cis-eQTL panels from 14 GTEx v8 brain tissues and whole blood( 39 ), enabling systematic assessment of tissue-specific expression–trait associations across multiple neuroanatomical regions. To further evaluate the contribution of cell type–specific regulatory mechanisms, we additionally incorporated cis-eQTL weights from nine neural cell populations reported by Bryois et al.( 41 ). These datasets included excitatory and inhibitory neurons, astrocytes, oligodendrocytes, microglia, OPCs, endothelial cells, and other key neurovascular cell types. All SMR and HEIDI analyses were performed using default settings. Multiple testing correction was applied using the Benjamini–Hochberg FDR, and genes passing both SMR significance and HEIDI filtering were retained as putative causal candidates for insomnia and ALPS phenotypes. Multi-Trait Analysis of GWAS (MTAG) To increase statistical power and improve the discovery of shared and trait-specific association signals, we applied MTAG, an extension of univariate GWAS that jointly analyzes summary statistics from genetically correlated traits( 42 ). MTAG models the vector of GWAS Z-scores across traits under a multivariate normal framework that accommodates genetic correlation and explicitly corrects for sample overlap, thereby enabling each trait to “borrow strength” from others to enhance effective association power. We followed the standard MTAG workflow, using harmonized GWAS summary statistics for insomnia and the seven ALPS phenotypes as input. MTAG was run using default parameter settings, and standard error corrections were applied to account for overlapping samples across input GWAS datasets. This procedure yielded MTAG-enhanced GWAS summary statistics for each ALPS phenotype conditioned on shared genetic information with insomnia. Analyses were performed in Python using the publicly available MTAG software released by Turley et al.( 42 ). All MTAG outputs underwent standard quality checks and were subsequently used in downstream gene-mapping, tissue- and cell-type enrichment, and transcriptomic integration analyses. Tissue- and cell-type enrichment analysis using Phenotype–Cell–Gene Association (PCGA) To systematically evaluate the contribution of tissues and cell types to insomnia and GS function, we applied the PCGA framework( 43 ). PCGA builds on the Driver tissue and gene Estimation (DESE) methodology to jointly infer trait-relevant tissues, cell types, and putative susceptibility genes by integrating GWAS summary statistics with large-scale bulk and single-cell transcriptomic resources. The PCGA reference compendium includes expression profiles from 54 human tissues, 2,214 human cell types, and 4,384 mouse cell types, allowing high-resolution inference across multiple biological hierarchies. In this study, we first generated MTAG-enhanced GWAS summary statistics for insomnia and each of the seven ALPS phenotypes. These MTAG-META results were then used as input to PCGA, which models GWAS association patterns jointly with tissue- and cell-type–specific expression signatures to identify enriched biological compartments. Through its hierarchical DESE-based inference, PCGA estimates the tissues, cell types, and gene sets most likely to underlie trait-associated genetic signals. All analyses were performed using the default PCGA workflow, reference datasets, and significance evaluation procedures. Enriched tissues, cell populations, and inferred susceptibility genes were retained as candidates contributing to shared insomnia–ALPS biology. Spatially resolved genetic mapping using gsMap We applied genetically informed gsMap to integrate GWAS summary statistics with ST data and identify spatially localized cell populations and genes associated with insomnia, AD, and the seven ALPS phenotypes( 44 ). gsMap first uses a graph neural network (GNN) to learn low-dimensional embeddings for each ST spot by jointly modeling gene expression profiles and spatial coordinates. Spots are then grouped into homogeneous microdomains based on cosine similarity, and gene-specificity scores (GSS)—quantifying the spatial specificity of each gene—are computed. GSS values are mapped to nearby SNPs (± 50 kb around the transcription start site) or to distal variants using external chromatin-based enhancer annotations (e.g., ABC model). For each ST spot, gsMap applies stratified LD score regression (S-LDSC) to test whether SNPs with high GSS values are enriched for trait heritability. To evaluate regional associations (e.g., anatomical zones or microdomains), gsMap aggregates spot-level p-values using the Cauchy combination test to generate p_cauchy statistics. Additional metrics—including p_median, the regional median GSS, and the Pearson correlation coefficient (PCC) between GSS and local expression—provide complementary measures of spatial association strength and specificity. In this study, we analyzed GWAS summary statistics for insomnia, AD, and all seven ALPS phenotypes using two complementary ST datasets: (i) human postmortem ITC Visium data (Visium_SPG_AD) and (ii) E16.5 mouse brain MOSTA sections, reflecting early neuroglial and neurovascular development. Both spot-level and region-level analyses were performed to identify spatial hotspots, trait-associated microdomains, and candidate genes with anatomically constrained effects. RESULTS Polygenic analyses reveal non-random genetic overlap between insomnia and ALPS phenotypes To assess the genetic relationship between insomnia and ALPS, we first constructed cross-trait conditional Q–Q plots. Conditioning on ALPS, insomnia displayed a marked upward deviation from the null among SNPs with increasing ALPS significance (e.g., p_ALPS < 10⁻³), and the reciprocal pattern was likewise observed. This bidirectional deviation indicates substantial polygenic enrichment between the two traits beyond chance expectations (Supplementary Fig. 1–8). Polygenic modeling provided additional support for this architecture. Although insomnia showed lower SNP-based heritability (h² = 0.080) than ALPS phenotypes (h² = 0.135–0.189) (Fig. 2 A), the two traits exhibited comparable proportions of non-zero-effect SNPs (π) and similar effect size variances (σ²β). Significant cross-trait conditional enrichment persisted, suggesting the presence of overlapping sets of risk-influencing variants (Supplementary Table 2). To determine whether this enrichment reflects genuine shared genetic architecture rather than correlated noise, we applied bivariate MiXeR and GPA. Genome-wide genetic correlations were near zero (rg_MiXeR = − 0.038 to 0.003), yet both frameworks indicated modest but detectable overlap in causal variants (MiXeR Dice coefficients: 17.64%–42.37%; GPA π₁₁: 6.44%–9.07%) (Fig. 2 B, C, D). Nearly half of the shared SNPs (45.7%–49.7%) exhibited opposite effect directions, consistent with antagonistic pleiotropy that may obscure global genetic correlations despite localized sharing. Importantly, all joint models significantly outperformed independence models (χ² > 6300, p < 10⁻³⁰⁰), demonstrating robust, non-random, region-specific genetic overlap between insomnia and ALPS phenotypes (Supplementary Tables 3–4). Insert Fig. 2 Local genetic correlation and colocalization analyses identify 17q21.31 as a core locus shared between insomnia and ALPS phenotypes Local genetic correlation analysis using LAVA identified several regions with significant local genetic correlation (FDR < 0.05), among which the chr17:42,348,004–43,460,500 interval (17q21.31) emerged as the most prominent. This region showed strong positive local correlations between insomnia and both Left_ALPS (ρ = 0.547, FDR = 0.008) and Mean_ALPS (ρ = 0.564, FDR = 0.010), indicating focal convergence of genetic effects despite negligible genome-wide correlation. Bayesian colocalization using GWAS-PW further supported shared architecture at this locus: the posterior probability that insomnia and ALPS share a single causal variant (Model 3) was extremely high (PPA₃ > 0.99), providing strong evidence for true colocalization (Fig. 3 A; Supplementary Table 5). These findings collectively implicate 17q21.31 as a central pleiotropic hotspot influencing both insomnia susceptibility and GS function. Insert Fig. 3 ConjFDR and Bayesian colocalization identify two independent shared causal signals within 17q21.31 Using the conjFDR framework (FDR < 0.05), we identified 16 pleiotropic variants jointly associated with insomnia and at least one ALPS phenotype, among which two mapped to the 17q21.31 locus. The first signal, rs71373536—located 259 bp downstream of ACBD4 —showed significant cross-trait enrichment for both insomnia–Left_ALPS (conjFDR = 0.004) and insomnia–Mean_ALPS (conjFDR = 0.004), with concordant effect directions across traits (Fig. 3 B, C). The second signal, rs1044977, a missense variant within HEXIM1 (CADD = 18.79), was specifically detected for insomnia–Right_ALPS (conjFDR = 0.020) (Fig. 4 A; Supplementary Table 6). Bayesian colocalization further validated these findings. Within ± 100 kb windows, rs71373536 exhibited strong evidence for a shared causal variant between insomnia and ALPS phenotypes (PP.H4 = 0.929–0.936), whereas rs1044977 showed similarly high colocalization support (PP.H4 = 0.861). Both signals reside within the chr17:43.2–43.3 Mb interval, overlapping the region independently highlighted by GWAS-PW (Supplementary Table 6). Fine-mapping identifies multiple high-confidence causal variants, revealing the complex LD structure of 17q21.31 To further evaluate the number and confidence of putative causal variants underlying the two pleiotropic signals identified by conjFDR (rs1044977 and rs71373536), we performed fine-mapping within ± 100 kb of each lead SNP. In the rs1044977 region, both FINEMAP and SuSiE converged on a model supporting five independent causal variants. Among these, rs9894577 (chr17:43,223,292) was consistently assigned the highest posterior probability (PP ≈ 1.0) and represents the strongest causal candidate, located near the ACBD4–HEXIM1–PLCD3 gene cluster. Although rs1044977 itself was not included in the top credible set, it is in very strong LD with rs9894577 (r² > 0.95) and carries a high CADD score, suggesting that it may act as a functional proxy or contribute directly to the causal signal. Fine-mapping in the rs71373536 region similarly supported five independent causal components. The variant rs6503419 (chr17:43,149,988) emerged as the dominant causal signal (PP = 1.0), residing near the DCAKD–NMT1–PLCD3 interval. rs71373536 is strongly correlated with rs6503419 (r² > 0.8), indicating that these two SNPs likely tag the same underlying causal configuration (Fig. 4 B; Supplementary Table 7). Together, these results highlight that the 17q21.31 locus contains at least two tightly linked but potentially functionally distinct shared causal signals, consistent with the region’s well-recognized extended LD structure and complex haplotype architecture. Insert Fig. 4 Gene prioritization and causal inference identify key shared genes within 17q21.31 To functionally resolve the shared genetic signals at 17q21.31, we performed gene-level fine-mapping using FOCUS. Two genes— HEXIM1 and EFTUD2 —emerged with high posterior inclusion probabilities across multiple brain tissues. HEXIM1 showed consistently strong support (PIP > 0.94 across cerebellar hemisphere, anterior cingulate cortex, and whole-blood models) and exhibited a significant negative TWAS association with insomnia (Z = − 7.43), indicating that reduced expression increases disease risk. Its missense variant (rs1044977, CADD = 18.79) further suggests functional perturbation, nominating HEXIM1 as a key mediator linking sleep disturbance with neurodegenerative pathways. EFTUD2 also showed strong evidence (PIP > 0.95 across frontal cortex, hippocampus, and dorsolateral prefrontal cortex), with expression positively associated with insomnia risk (Z = 7.92) (Supplementary Table 8). To complement fine-mapping, we applied S-PrediXcan focusing on genes annotated to colocalized lead SNPs within 17q21.31 (Supplementary Table 9). In whole blood, ACBD4 expression was significantly associated with insomnia (Z = 4.02, FDR = 0.0003), and also with Right_ALPS (Z = 3.19, FDR = 0.007) and Mean_ALPS (Z = 2.97, FDR = 0.015) (Fig. 5 A; Supplementary Table 10). SMR analysis confirmed this pattern: ACBD4 (top SNP: rs2291447) showed significant positive causal effects on insomnia and multiple ALPS phenotypes (b_SMR > 0, FDR < 0.05), suggesting that ACBD4 -mediated processes in peripheral blood may contribute to the shared genetic architecture between insomnia and GS function (Fig. 5 B, C; Supplementary Table 11). Using S-MultiXcan across 13 GTEx brain tissues, three genes— HEXIM1 , ACBD4 , and MAPT —displayed significant associations for insomnia and at least one ALPS phenotype (FDR < 0.05). HEXIM1 remained the most strongly supported gene, showing consistent negative associations with insomnia (Z_mean = − 6.77, FDR = 1.56×10⁻¹⁵) and with Left_ALPS, Mean_ALPS, and Right_ALPS (Z_mean = − 2.63 to − 2.81), with strongest signals arising in basal ganglia regions (Putamen and Caudate), in agreement with FOCUS results (Fig. 5 A; Supplementary Table 12). ACBD4 showed a significant positive association with insomnia (Z_mean = 3.06, FDR = 6.43×10⁻¹¹), with its strongest tissue signal localized to the nucleus accumbens (Brain_Nucleus_accumbens_basal_ganglia, p_i_best = 6.05×10 − 6 ). ACBD4 expression was also positively associated with Mean_ALPS and Right_ALPS (Z_mean = 2.49–2.60, FDR = 0.045–0.048), with peak associations observed in the hypothalamus (p_i_best = 8.42×10⁻⁴–8.66×10⁻⁴), consistent with S-PrediXcan results (Fig. 5 A; Supplementary Table 12). MAPT exhibited a significant positive association with insomnia (Z_mean = 1.63, FDR = 0.0071), with the strongest signal observed in the cervical spinal cord. In contrast, its association with Left_ALPS was negative (Z_mean = − 1.04, FDR = 0.016), with the peak association arising from the caudate nucleus (Fig. 5 A; Supplementary Table 12). SMR analyses across 13 brain tissues and nine neural cell types further delineated the tissue- and cell-specific architecture of the shared loci. ACBD4 showed consistent positive causal effects on insomnia and multiple ALPS phenotypes across several central nervous system regions, including the cerebellum, cerebellar hemisphere, and cervical spinal cord (b_SMR > 0, FDR < 0.05). These findings align with the TWAS results and indicate that higher ACBD4 expression in these regions is associated with increased insomnia risk as well as elevated Right_ALPS and Mean_ALPS metrics (Fig. 5 B, C; Supplementary Table 13). In contrast, MAPT expression in astrocytes displayed significant negative causal effects on aALPS, mALPS, Mean_ALPS, pALPS, Right_ALPS and tALPS (b_SMR < 0, FDR < 0.05). This pattern is consistent with the S-MultiXcan results and indicates that astrocytic MAPT expression is associated with reduced glymphatic-related diffusivity across multiple ALPS measures (Fig. 5 D; Supplementary Table 14). Spatial, tissue, and cell-type enrichment analyses reveal anatomically distributed but phenotype-specific substrates of insomnia–GS pleiotropy Integrative genetic analyses identified HEXIM1 and ACBD4 at the 17q21.31 locus as shared molecular hubs for insomnia and ALPS phenotypes, and further revealed significant associations with MAPT —the core pathogenic gene of AD. Because 17q21.31 represents one of the most intensively characterized AD risk regions and MAPT -driven tau pathology is a hallmark of AD, the presence of convergent signals at this locus suggested that pleiotropic variants influencing insomnia–ALPS coupling may also intersect with AD-related pathogenic processes. Given that ALPS phenotypes index GS function, including clearance of neurotoxic proteins such as Aβ and pTau, we sought to spatially contextualize these shared genetic signals within an AD-relevant microenvironment. To this end, we analyzed ST data from the ITC of AD donors using gsMap to map the spatial and cellular distribution of trait-associated genes. Applying gsMap to the ITC dataset revealed that aALPS exhibited exceptionally strong enrichment across multiple neuronal and glial populations, with the most pronounced signals in excitatory neurons (p_cauchy = 9.60×10⁻⁶⁴) and inhibitory neurons (p_cauchy = 6.59×10⁻¹⁸), alongside robust enrichment in astrocytes, microglia, and OPCs (all p_cauchy < 5×10⁻⁷). Notably, aALPS-associated genetic signals localized not only to Aβ and pTau deposition cores but also to surrounding periplaque regions (n_Aβ and n_pTau, denoting regions “next to Aβ/pTau deposits” in the original ST annotation), suggesting that these variants may influence broader microenvironmental responses to protein aggregation rather than the deposition process itself (Supplementary Table 15). Other ALPS phenotypes displayed more anatomically circumscribed patterns. Left_ALPS (p_cauchy = 1.39×10⁻¹⁴, p_median = 0.84) and Mean_ALPS (p_cauchy = 7.74×10⁻⁶, p_median = 0.84) were most strongly enriched within Aβ-positive regions, whereas tALPS showed peak enrichment in astrocytes (p_cauchy = 8.94×10⁻¹⁰, p_median = 0.81). In contrast, insomnia exhibited no significant enrichment in any spatial domain or cell type, while AD GWAS signals were highly enriched in microglia and multiple Aβ/pTau pathological zones (all p_cauchy < 5×10⁻⁷), providing an important reference for interpreting ALPS–AD convergence (Supplementary Table 15). To validate these observations and further delineate the shared genetic architecture, we performed a cross-trait MTAG meta-analysis combining insomnia with each ALPS phenotype, followed by tissue and cell-type enrichment using PCGA. Shared genes identified by MTAG showed widespread enrichment across diverse brain regions—including hippocampus, BA9, caudate, putamen, nucleus accumbens, BA24, amygdala, hypothalamus, cerebellum, cervical spinal cord, and substantia nigra—and across major neural lineages, including excitatory and inhibitory neurons, astrocytes, microglia, oligodendrocytes and OPCs, vascular–pia populations, endothelial cells, and retinal ganglion cells (P_adj < 5×10⁻⁷). These results were highly consistent across insomnia–aALPS, –Left_ALPS, –Mean_ALPS, and –tALPS shared gene sets, supporting a distributed neuroanatomical and multi-lineage cellular architecture underlying insomnia–GS pleiotropy (Fig. 5 E; Supplementary Table 16–17). Cross-species gsMap analysis using E16.5 mouse MOSTA sections further corroborated these spatial enrichment patterns. Shared insomnia–ALPS genes were significantly enriched in the spinal cord, brain, and choroid plexus (p_cauchy < 5×10⁻⁷), key tissues involved in early neuroglial maturation and the anatomical substrates that later support GS function (Supplementary Table 18). To further resolve gene-level spatial signatures, we profiled three key genes. AQP4 , a canonical structural component of the GS pathway, showed strong enrichment in Ab/pTau-adjacent regions (enrichment score = 1.88, FDR < 0.01) and in astrocytes (score = 2.56, FDR 0, P_bon < 0.05) except aALPS, for which the correlation was negative (r < 0, P_bon < 0.05). This opposite direction of association relative to other ALPS measures highlights a marked regional specificity in the physiological processes captured by different ALPS indices (Fig. 5 F; Supplementary Table 19). MAPT was strongly enriched in excitatory neurons (enrichment score = 1.94, FDR < 0.01) as well as in regions adjacent to both Aβ and pTau pathology (score = 1.84, FDR < 0.01). Within excitatory neurons, MAPT expression showed negative correlations with all ALPS phenotypes (r < 0, P_bon 0, P_bon < 0.05) (Fig. 5 F; Supplementary Table 19). HEXIM1 showed enrichment in excitatory neurons (score = 1.74, FDR < 0.01) and Aβ regions (score = 1.74, FDR < 0.01), and consistently negative correlations with all ALPS phenotypes (r < 0, P_bon < 0.05), fully concordant with TWAS and SMR results (Fig. 5 F; Supplementary Table 19). Together, these spatial and cellular analyses demonstrate that insomnia–ALPS shared genes—particularly those at 17q21.31—are distributed across multiple neuronal and glial compartments, yet exhibit sharply phenotype-dependent and region-specific associations with GS function. These findings highlight that genetic effects on GS are highly context-dependent, varying across spatial microenvironments, cell types, and AD-related pathological regions. Insert Fig. 5 DISCUSSION By integrating large-scale GWAS of insomnia with hemispheric and region-specific ALPS phenotypes, this study provides a systematic genetic framework linking sleep disturbance to GS function. We demonstrate that: (i) insomnia and ALPS exhibit significant polygenic enrichment despite negligible genome-wide genetic correlation; (ii) the extended LD block at 17q21.31 constitutes a major shared susceptibility locus; (iii) multiple genes within this region, including HEXIM1 , ACBD4 , EFTUD2 , and MAPT , show convergent associations across statistical layers; (iv) shared genetic signals derived from MTAG meta-analysis display broad yet structured enrichment across diverse brain regions and neuroglial cell populations; and (v) The astrocyte-specific eQTL effect of MAPT , together with the known role of glia in glymphatic regulation, suggests a potential glial pathway through which insomnia may influence perivascular clearance. Together, these findings define a multilayer genetic link between insomnia traits and GS function, extending molecular hypotheses connecting sleep disruption, glial dysfunction, and vulnerability to neurodegenerative processes. Targeted rather than genome-wide genetic coupling between insomnia and GS Although insomnia and ALPS showed negligible genome-wide genetic correlation, conjFDR revealed robust polygenic overlap. This pattern indicates that the two traits do not share a broadly diffuse genetic architecture, but instead converge through a targeted subset of molecular pathways. Mixture-model results, showing a large fraction of variants with opposite effects across traits, further support the view that GS-related biology forms a distinct yet functionally relevant component of the broader insomnia genetic landscape. These findings help explain why prior GWAS studies of sleep and neurological traits detected limited overlap( 8 , 45 ): shared biology exists, but its genetic footprint is localized rather than global, necessitating enrichment-based analytic strategies. 17q21.31 as a shared genetic hub linking insomnia, GS, and neurodegeneration A central contribution of this work is the identification of 17q21.31 as the major locus jointly influencing insomnia and GS function. This region spans the MAPT H1/H2 inversion haplotype, characterized by extended LD, dense regulatory elements, and broad pleiotropic effects in AD, Parkinson’s disease, and frontotemporal dementia( 46 – 48 ). Because of the inversion, statistical associations largely reflect haplotype-level architecture rather than resolvable independent signals. Within this constraint, fine-mapping, TWAS, and SMR provide convergent functional prioritization rather than definitive causal separation. Fine-mapping highlighted HEXIM1 and EFTUD2 as top posterior candidates, while TWAS and SMR pointed to ACBD4 and MAPT . These findings suggest that multiple regulatory elements embedded in the 17q21.31 haplotype may contribute to insomnia–GS biology, even though the underlying genetic architecture remains haplotype-driven. Mechanistically, these prioritized genes converge on pathways involving transcriptional regulation ( HEXIM1 )( 49 ), peroxisomal and lipid metabolism ( ACBD4 )( 50 ), RNA splicing ( EFTUD2 )( 51 ), and tau-related astroglial function ( MAPT ). The astrocyte-specific negative causal effect of MAPT on ALPS indices is particularly notable, highlighting a glial pathway through which insomnia may impair perivascular clearance. Prior ALPS GWAS examining neurodegenerative biomarkers have identified a shared MAPT locus between CSF p-tau and ALPS measures (rs7521; condFDR = 0.0071), indicating that tau-related processes and GS function converge at the same 17q21.31 haplotype( 14 ). This aligns with experimental evidence that tau pathology disrupts astrocytic endfeet and aquaporin-4 polarity( 52 ), thereby compromising perivascular solute transport. Importantly, external evidence from prior ALPS GWAS further reinforces the role of 17q21.31 in sleep–GS coupling( 14 ). In the ALPS dataset, independent significant variants within the MAPT inversion block (rs62061734 and rs1991556) were previously reported as genome-wide significant loci for sleep duration in two large-scale GWAS (N ≈ 1.3 million) conducted by Jansen et al.( 53 ) and Doherty et al.( 54 ) These SNPs map to the same region implicated in our insomnia–GS analyses, indicating that sleep duration, insomnia liability, and GS-related perivascular clearance share convergent regulatory architecture within the 17q21.31 haplotype. Together, these findings position 17q21.31 as a biological nexus integrating sleep regulation, GS function, and vulnerability to neurodegenerative processes. Distributed but directionally divergent architecture across brain regions and cell types A notable aspect of our findings is that the shared genetic architecture between insomnia and GS function does not manifest as regional specificity. PCGA analyses revealed that shared genes were broadly enriched across multiple brain regions—including the hippocampus, prefrontal cortex, basal ganglia, hypothalamus, cerebellum, and spinal cord—and across diverse neuronal and glial lineages. This pattern indicates a distributed neuroanatomical and cellular footprint, rather than enrichment restricted to any single GS-relevant compartment. However, fine-grained analyses revealed a different organizational principle. TWAS, SMR, and gsMap indicated that the direction of genetic effects frequently diverged across ALPS phenotypes and anatomical contexts, even when involving the same gene. MAPT provides a clear example: although consistently showing negative associations with most ALPS indices across methods, its effect became positive for aALPS specifically within excitatory neurons. This illustrates that a single GS-related gene can exert opposite influences depending on local neurovascular and cellular environments. AQP4 , a canonical astrocytic water-channel gene( 55 ), showed a similar pattern in our gsMap results—positive associations with most hemispheric and posterior/middle ALPS indices but a negative association with aALPS. Although AQP4 is not among the core shared loci between insomnia and GS function, its well-established role in perivascular flux supports the biological plausibility of such context-dependent directional reversals. Together, these results support a distributed-but-directionally-divergent model of insomnia–GS genetic coupling. Shared genes are broadly expressed across the brain, but their functional effects on perivascular clearance are shaped by local neurovascular and cellular context. This framework aligns with the compartmentalized nature of GS architecture( 56 ), where regional differences in perivascular spacing, arterial pulsatility, and astrocytic polarity can invert genetic effects even when the same molecular pathways are involved. Thus, instead of classical spatial specificity, the insomnia–GS interface is characterized by context-dependent functional heterogeneity across the brain. Cross-species ST mapping highlights developmental neuroglial substrates To refine the anatomical context of shared pathways, we leveraged mouse gsMap using MTAG-derived shared genes. The use of E16.5 mouse ST data is biologically justified because this developmental stage represents a critical window during which astrocytes, radial glia, and neurovascular scaffolding are actively established—cellular and structural precursors that later form the core architecture of glymphatic pathways( 57 ). Importantly, gene-expression gradients at this stage are highly conserved across species and often predict adult regional specialization. Shared insomnia–ALPS genes showed selective localization within cortical and subcortical territories characterized by high metabolic demand and dense neuroglial interactions, mirroring the adult brain regions implicated in our PCGA and gsMap analyses. This concordance suggests that the molecular interface between sleep regulation and GS function may be rooted in early neurodevelopmental programs that specify astroglial polarity, perivascular organization, and neurovascular coupling. In this framework, embryonic spatial patterning provides the foundation upon which adult GS efficiency and sleep-related clearance vulnerability are built, offering a developmental perspective on the emergence of insomnia–GS genetic convergence. Mechanistic insights into insomnia–GS interactions Cell-type–specific TWAS and SMR analyses identified astrocytes as a convergent interface linking insomnia to GS function. Astrocytic MAPT effects support mechanisms involving altered endfoot polarity, AQP4 mislocalization, or tau-driven disruption of neurovascular coupling( 52 ). HEXIM1 ( 49 ) and ACBD4 ( 50 ) point toward transcriptional and metabolic pathways influencing astrocyte and endothelial contributions to perivascular fluid dynamics. Together, these results outline a model in which insomnia modifies glia-regulated clearance pathways, influencing regional susceptibility to impaired glymphatic transport and potentially facilitating downstream toxic protein accumulation. Limitations and future directions Several limitations warrant consideration. First, although ALPS indices provide genetically informative and scalable proxies of GS function, they capture only perivascular diffusivity and do not encompass the full GS pathway—particularly meningeal lymphatic drainage( 58 )—and no gold-standard in vivo metric of GS function currently exists. Second, the insomnia GWAS used in this study is based on self-reported symptoms in UK Biobank, which introduces two layers of unavoidable uncertainty: (i) the primary GWAS does not explicitly exclude individuals with neurodegenerative disorders, raising the possibility that a subset of cases reflects sleep disturbance secondary to early or established neurodegeneration rather than primary insomnia; and (ii) insomnia was defined using questionnaire-based phenotyping rather than clinical diagnosis, which increases heterogeneity and may attenuate trait-specific genetic signals. Although the original GWAS included sensitivity analyses excluding selected chronic and psychiatric illnesses, comprehensive removal of neurodegenerative cases is not feasible using publicly available summary statistics. Third, ST mapping analyses were limited by the lack of ST data from individuals with insomnia. Finally, although we identified convergent genetic pathways linking insomnia and GS function, causal directionality among insomnia, GS dysfunction, and neurodegenerative processes cannot be inferred from cross-sectional GWAS and requires longitudinal and experimental confirmation. Future studies combining dynamic sleep recordings, multimodal neuroimaging, CSF biomarkers, and cell-type–specific perturbations will be essential for elucidating how sleep behavior, glial physiology, and clearance mechanisms interact over time. Such efforts may help identify therapeutic strategies targeting sleep or GS pathways to mitigate neurodegenerative risk. Declarations DATA AVAILABILITY Insomnia GWAS: publicly available at the GWAS Catalog (GCST007387), https://www.ebi.ac.uk/gwas/studies/GCST007387. ALPS GWAS: available via the GWAS Catalog under publications 39823331 and 39805841. https://www.ebi.ac.uk/gwas/publications/39823331. https://www.ebi.ac.uk/gwas/publications/39805841. GTEx v8 eQTLs (for SMR and TWAS): https://yanglab.westlake.edu.cn/software/smr/#eQTLsummarydata. Cell-type–specific cis-eQTLs: https://zenodo.org/records/7276971. Cell type–specific prediction models for scPrediXcan: https://zenodo.org/records/14346661. 1000 Genomes Phase 3 reference (European): available through standard LD reference panels used by MiXeR, LAVA, and SMR MAGMA gene location files: https://ctg.cncr.nl/software/MAGMA/aux_files/NCBI37.3.zip. gsMap spatial framework: https://github.com/LieberInstitute/Visium_SPG_AD. All QC-filtered GWAS files (Insomnia, ALPS), the harmonized datasets used for MiXeR, LAVA, GPA, and PleioFDR analyses, and the processed spatial transcriptomic matrices used for gsMap have been deposited in Figshare and are accessible at: https://doi.org/10.6084/m9.figshare.30751190. All data used in the manuscript are publicly available, and no individual-level confidential data were used. SOFTWARE USED All analyses were performed using publicly available software. The key tools include: GPA (v1.2.1), https://dongjunchung.github.io/GPA/ MiXeR (v2.2.1), https://github.com/precimed/mixer LAVA (v0.1.0), https://github.com/josefin-werme/LAVA GWAS-PW, https://github.com/joepickrell/gwas-pw PleioFDR (condFDR & conjFDR), https://github.com/precimed/pleiofdr MTAG, https://github.com/JonJala/mtag Fine-mapping, colocalization, and integrative functional analyses COLOC, https://github.com/chr1swallace/coloc echolocatoR (v1.1+), https://rajlabmssm.github.io/echolocatoR/index.html FOCUS, https://github.com/mancusolab/ma-focus/wiki SMR & HEIDI, https://yanglab.westlake.edu.cn/software/smr/#SMR&HEIDIanalysis MetaXcan, https://github.com/hakyimlab/MetaXcan scPrediXcan, https://github.com/hakyimlab/scPrediXcan Cell-type and spatial-functional annotation analyses PCGA, https://pmglab.top/pcga/#/infer_cell GsMap, https://yanglab.westlake.edu.cn/gsmap/document/software clusterProfiler (v4.10+), http://bioconductor.org/packages/release/bioc/html/clusterProfiler.html aPEAR, https://github.com/kerseviciute/aPEAR General genomic annotation tools FUMA, https://fuma.ctglab.nl/ All additional R and Python packages used in preprocessing, visualization, and statistical analyses are standard and listed within the GitHub repository described in the “Code Availability” section. CODE AVAILABILITY All scripts used for preprocessing, QC, statistical analyses, and figure generation are based on the publicly available software and repositories cited in the “Software Availability” section. The custom code used to reproduce the analytical workflow (including data harmonization, MiXeR fitting pipelines, LAVA block-level analyses, multi-method fine-mapping, conditional FDR procedures, TWAS/MetaXcan models, cell-type mapping, and gsMap integration) has been deposited in GitHub: https://github.com/junkman666/Insomnia.git ACKNOWLEDGEMENTS We thank the research teams who generated and openly shared the GWAS, eQTL, cell-type–specific, and ST datasets that made this study possible. We are grateful to the developers and maintainers of the publicly available tools used in this work, including MiXeR, LAVA, GPA, PleioFDR, GWAS-PW, COLOC, FOCUS, MetaXcan, scPrediXcan, PCGA, GsMap, and associated computational pipelines. AUTHOR CONTRIBUTIONS J.S., Y.S. and C.X. designed and supervised the study. J.S. performed data preprocessing, statistical analyses, and integrative genomic modeling. Y.S. and C.X. contributed to methodological refinement, result interpretation, and quality control. All authors drafted, revised, and approved the final manuscript. FUNDING This work was supported by Suzhou Science and Technology Bureau Basic research on medical applications -Research on innovative medical applications Project (SKY2023101). COMPETING INTERESTS The authors declare no competing interests. ETHICS APPROVAL This study used only publicly available, de-identified summary-level genomic and transcriptomic data. No new human participant recruitment, intervention, or identifiable information was involved. Therefore, ethics approval was not required. INFORMED CONSENT All datasets analyzed in this study were obtained from public repositories with existing ethics approval and participant informed consent procedures. No new consent was required for the reuse of anonymized summary data. References Adamantidis, A.R., de Lecea, L.: Sleep and the hypothalamus. Science. 382 , 405–412 (2023) Zou, Q., Zou, G., Wang, S., Wang, Y., Xu, J., Long, Y., et al.: Cortical hierarchy underlying homeostatic sleep pressure alleviation. Nat. Commun. 16 , 10014 (2025) Chaput, J.-P., McHill, A.W., Cox, R.C., Broussard, J.L., Dutil, C., da Costa, B.G.G., et al.: The role of insufficient sleep and circadian misalignment in obesity. Nat. Rev. Endocrinol. 19 , 82–97 (2023) Zhang, X., Yin, J., Sun, X., Qu, Z., Zhang, J., Zhang, H.: The association between insomnia and cognitive decline: A scoping review. Sleep. Med. 124 , 540–550 (2024) Perlis, M.L., Posner, D., Riemann, D., Bastien, C.H., Teel, J., Thase, M.: Insomnia. Lancet Lond. Engl. 400 , 1047–1060 (2022) Blanco-García, F.J., López-López, J.A., Ordoñana, J.R., Madrid-Valero, J.J.: Genetic association between sleep quality, insomnia, and psychological distress: a systematic review and meta-analysis. Sleep. Med. Rev. 83 , 102149 (2025) Madrid-Valero, J.J., Gregory, A.M.: Behaviour genetics and sleep: A narrative review of the last decade of quantitative and molecular genetic research in humans. Sleep. Med. Rev. 69 , 101769 (2023) Sonti, S., Grant, S.F.A.: Leveraging genetic discoveries for sleep to determine causal relationships with common complex traits. Sleep. 45 , zsac180 (2022) Kozhemiako, N., Mita, C., Panagiotaropoulou, G.M., Lewis, K.J.S., Sofer, T., Purcell, S.M.: The Interplay Between Sleep and Mental Health: A Genetic Perspective. Biol. Psychiatry. 98 , 863–874 (2025) Chong, P.L.H., Garic, D., Shen, M.D., Lundgaard, I., Schwichtenberg, A.J.: Sleep, cerebrospinal fluid, and the glymphatic system: A systematic review. Sleep. Med. Rev. 61 , 101572 (2022) Reiter, R.J., Sharma, R., Cucielo, M.S., Tan, D.X., Rosales-Corral, S., Gancitano, G., de Almeida Chuffa, L.G.: Brain washing and neural health: role of age, sleep, and the cerebrospinal fluid melatonin rhythm. Cell. Mol. Life Sci. CMLS. 80 , 88 (2023) Rasmussen, M.K., Mestre, H., Nedergaard, M.: Fluid transport in the brain. Physiol. Rev. 102 , 1025–1151 (2022) Iliff, J.J., Wang, M., Liao, Y., Plogg, B.A., Peng, W., Gundersen, G.A., et al.: A paravascular pathway facilitates CSF flow through the brain parenchyma and the clearance of interstitial solutes, including amyloid β. Sci. Transl Med. 4 , 147ra111 (2012) Huang, S.-Y., Ge, Y.-J., Ren, P., Wu, B.-S., Gong, W., Du, J., et al.: Genome-wide association study unravels mechanisms of brain glymphatic activity. Nat. Commun. 16 , 626 (2025) Ran, L., Fang, Y., Cheng, C., He, Y., Shao, Z., Kong, Y., et al.: Genome-wide and phenome-wide studies provided insights into brain glymphatic system function and its clinical associations. Sci. Adv. 11 , eadr4606 (2025) Divecha, Y.A., Rampes, S., Tromp, S., Boyanova, S.T., Fleckney, A., Fidanboylu, M., Thomas, S.A.: The microcirculation, the blood-brain barrier, and the neurovascular unit in health and Alzheimer disease: The aberrant pericyte is a central player. Pharmacol. Rev. 77 , 100052 (2025) Lane, J.M., Jones, S.E., Dashti, H.S., Wood, A.R., Aragam, K.G., van Hees, V.T., et al.: Biological and clinical insights from genetics of insomnia symptoms. Nat. Genet. 51 , 387–393 (2019) Kwon, S.H., Parthiban, S., Tippani, M., Divecha, H.R., Eagles, N.J., Lobana, J.S., et al.: Influence of alzheimer’s disease related neuropathology on local microenvironment gene expression in the human inferior temporal cortex. GEN. Biotechnol. 2 , 399–417 (2023) Chen, A., Liao, S., Cheng, M., Ma, K., Wu, L., Lai, Y., et al.: Spatiotemporal transcriptomic atlas of mouse organogenesis using DNA nanoball-patterned arrays. Cell. 185 , 1777–1792e21 (2022) Holland, D., Frei, O., Desikan, R., Fan, C.-C., Shadrin, A.A., Smeland, O.B., et al.: Beyond SNP heritability: Polygenicity and discoverability of phenotypes estimated with a univariate Gaussian mixture model. PLoS Genet. 16 , e1008612 (2020) Frei, O., Holland, D., Smeland, O.B., Shadrin, A.A., Fan, C.C., Maeland, S., et al.: Bivariate causal mixture model quantifies polygenic overlap between complex traits beyond genetic correlation. Nat. Commun. 10 , 2417 (2019) Frei, O., Hindley, G., Shadrin, A.A., van der Meer, D., Akdeniz, B.C., Hagen, E., et al.: Improved functional mapping of complex trait heritability with GSA-MiXeR implicates biologically specific gene sets. Nat. Genet. 56 , 1310–1318 (2024) Chung, D., Yang, C., Li, C., Gelernter, J., Zhao, H.: GPA: a statistical approach to prioritizing GWAS results by integrating pleiotropy and annotation ((H. Tang, editor)). PLoS Genet 10: e1004787. (2014) Werme, J., van der Sluis, S., Posthuma, D., de Leeuw, C.A.: An integrated framework for local genetic correlation analysis. Nat. Genet. 54 , 274–282 (2022) Zeng, R., Jiang, R., Huang, W., Wang, J., Zhang, L., Ma, Y., et al.: Dissecting shared genetic architecture between obesity and multiple sclerosis. EBioMedicine. 93 , 104647 (2023) Smeland, O.B., Frei, O., Shadrin, A., O’Connell, K., Fan, C.-C., Bahrami, S., et al.: Discovery of shared genomic loci using the conditional false discovery rate approach. Hum. Genet. 139 , 85–94 (2020) Andreassen, O.A., Thompson, W.K., Schork, A.J., Ripke, S., Mattingsdal, M., Kelsoe, J.R., et al.: Improved detection of common variants associated with schizophrenia and bipolar disorder using pleiotropy-informed conditional false discovery rate. PLoS Genet. 9 , e1003455 (2013) Wallace, C.: A more accurate method for colocalisation analysis allowing for multiple causal variants. PLoS Genet. 17 , e1009440 (2021) Wang, G., Sarkar, A., Carbonetto, P., Stephens, M.: A simple new approach to variable selection in regression, with application to genetic fine mapping. J. R Stat. Soc. Ser. B Stat. Methodol. 82 , 1273–1300 (2020) Benner, C., Spencer, C.C.A., Havulinna, A.S., Salomaa, V., Ripatti, S., Pirinen, M.: FINEMAP: efficient variable selection using summary data from genome-wide association studies. Bioinforma Oxf. Engl. 32 , 1493–1501 (2016) Schilder, B.M., Humphrey, J., Raj, T.: echolocatoR: an automated end-to-end statistical and functional genomic fine-mapping pipeline. Bioinforma Oxf. Engl. 38 , 536–539 (2022) Mancuso, N., Freund, M.K., Johnson, R., Shi, H., Kichaev, G., Gusev, A., Pasaniuc, B.: Probabilistic fine-mapping of transcriptome-wide association studies. Nat. Genet. 51 , 675–682 (2019) Berisa, T., Pickrell, J.K.: Approximately independent linkage disequilibrium blocks in human populations. Bioinforma Oxf. Engl. 32 , 283–285 (2016) Shi, H., Burch, K.S., Johnson, R., Freund, M.K., Kichaev, G., Mancuso, N., et al.: Localizing Components of Shared Transethnic Genetic Architecture of Complex Traits from GWAS Summary Data. Am. J. Hum. Genet. 106 , 805–817 (2020) Watanabe, K., Taskesen, E., van Bochoven, A., Posthuma, D.: Functional mapping and annotation of genetic associations with FUMA. Nat. Commun. 8 , 1826 (2017) Wang, K., Li, M., Hakonarson, H.: ANNOVAR: functional annotation of genetic variants from high-throughput sequencing data. Nucleic Acids Res. 38 , e164 (2010) Barbeira, A.N., Dickinson, S.P., Bonazzola, R., Zheng, J., Wheeler, H.E., Torres, J.M., et al.: Exploring the phenotypic consequences of tissue specific gene expression variation inferred from GWAS summary statistics. Nat. Commun. 9 , 1825 (2018) Barbeira, A.N., Pividori, M., Zheng, J., Wheeler, H.E., Nicolae, D.L., Im, H.K.: Integrating predicted transcriptome from multiple tissues improves association detection. PLoS Genet. 15 , e1007889 (2019) GTEx Consortium: The GTEx Consortium atlas of genetic regulatory effects across human tissues. Science. 369 , 1318–1330 (2020) Zhu, Z., Zhang, F., Hu, H., Bakshi, A., Robinson, M.R., Powell, J.E., et al.: Integration of summary data from GWAS and eQTL studies predicts complex trait gene targets. Nat. Genet. 48 , 481–487 (2016) Bryois, J., Calini, D., Macnair, W., Foo, L., Urich, E., Ortmann, W., et al.: Cell-type-specific cis-eQTLs in eight human brain cell types identify novel risk genes for psychiatric and neurological disorders. Nat. Neurosci. 25 , 1104–1112 (2022) 23andMe Research Team, Social Science Genetic Association Consortium, Turley, P., Walters, R.K., Maghzian, O., Okbay, A., et al.: Multi-trait analysis of genome-wide association summary statistics using MTAG. Nat. Genet. 50 , 229–237 (2018) Yang, Y., Chen, J., Zhao, X., Gong, F., Liu, R., Miao, J., et al.: Genetic analysis reveals the shared genetic architecture between breast cancer and atrial fibrillation. Front. Genet. 16 , 1450259 (2025) Song, L., Chen, W., Hou, J., Guo, M., Yang, J.: Spatially resolved mapping of cells associated with human complex traits. Nature. 641 , 932–941 (2025) Stefani, A., Antelmi, E., Arnaldi, D., Arnulf, I., During, E., Högl, B., et al.: From mechanisms to future therapy: a synopsis of isolated REM sleep behavior disorder as early synuclein-related disease. Mol. Neurodegener. 20 , 19 (2025) Goh, S., Thiyagarajan, L., Dudding-Byth, T., Pinese, M., Kirk, E.P.: A systematic review and pooled analysis of penetrance estimates of copy-number variants associated with neurodevelopment. Genet. Med. Off J. Am. Coll. Med. Genet. 27 , 101227 (2025) Pedicone, C., Weitzman, S.A., Renton, A.E., Goate, A.M.: Unraveling the complex role of MAPT-containing H1 and H2 haplotypes in neurodegenerative diseases. Mol. Neurodegener. 19 , 43 (2024) Shaw-Smith, C., Pittman, A.M., Willatt, L., Martin, H., Rickman, L., Gribble, S., et al.: Microdeletion encompassing MAPT at chromosome 17q21.3 is associated with developmental delay and learning disability. Nat. Genet. 38 , 1032–1037 (2006) Fujinaga, K., Huang, F., Peterlin, B.M.: P-TEFb: The master regulator of transcription elongation. Mol. Cell. 83 , 393–403 (2023) Costello, J.L., Koster, J., Silva, B.S.C., Worthy, H.L., Schrader, T.A., Hacker, C., et al.: Differential roles for ACBD4 and ACBD5 in peroxisome-ER interactions and lipid metabolism. J. Biol. Chem. 299 , 105013 (2023) Beauchamp, M.-C., Jerome-Majewska, L.A.: A protective role for EFTUD2 in the brain. Neuron. 112 , 3378–3380 (2024) Ishida, K., Yamada, K., Nishiyama, R., Hashimoto, T., Nishida, I., Abe, Y., et al.: Glymphatic system clears extracellular tau and protects from tau aggregation and neurodegeneration. J. Exp. Med. 219 , e20211275 (2022) Jansen, P.R., Watanabe, K., Stringer, S., Skene, N., Bryois, J., Hammerschlag, A.R., et al.: Genome-wide analysis of insomnia in 1,331,010 individuals identifies new risk loci and functional pathways. Nat. Genet. 51 , 394–403 (2019) Doherty, A., Smith-Byrne, K., Ferreira, T., Holmes, M.V., Holmes, C., Pulit, S.L., Lindgren, C.M.: GWAS identifies 14 loci for device-measured physical activity and sleep duration. Nat. Commun. 9 , 5257 (2018) Nagelhus, E.A., Ottersen, O.P.: Physiological roles of aquaporin-4 in brain. Physiol. Rev. 93 , 1543–1562 (2013) Pinho-Correia, L.M., McCullough, S.J.C., Ghanizada, H., Nedergaard, M., Rustenhoven, J., Da Mesquita, S.: CSF transport at the brain-meningeal border: effects on neurological health and disease. Lancet Neurol. 24 , 535–547 (2025) Li, X., Liu, G., Yang, L., Li, Z., Zhang, Z., Xu, Z., et al.: Decoding Cortical Glial Cell Development. Neurosci. Bull. 37 , 440–460 (2021) Keil, S.A., Jansson, D., Braun, M., Iliff, J.J.: Glymphatic dysfunction in Alzheimer’s disease: A critical appraisal. Science. 389 , eadv8269 (2025) Additional Declarations There is NO Competing Interest. Supplementary Files SupplementaryDatas.pdf Supplementary Datas SupplementaryFigures.docx Supplementary Figures Cite Share Download PDF Status: Under Review Version 1 posted You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-8585990","acceptedTermsAndConditions":true,"allowDirectSubmit":false,"archivedVersions":[],"articleType":"Article","associatedPublications":[],"authors":[{"id":594399825,"identity":"3e21247c-4293-4a63-9d6b-e191d845dd1f","order_by":0,"name":"Chenxu Xiao","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAAyklEQVRIiWNgGAWjYHAC9g8SPDVyINaBB0RqYWOwkDlmDNaSQLSWChvmxAYQkygtBsfPmD24kcOWPj/s8EOgLXZyug0EtEj2pKUbzjgjk7vxdpoBUEuysdkBAlr4JZgPSEv2sOVunJ0A0nIgcRshLWwSjA3Sf/8xpxvOTv9AnBagLcckJHiYE+Slc4i0BeiXZAMJnmOGG6RzCg4kGBDhF2CIGT4ARqW8/Oz0zR8+VNjJEdSC0AtWaUCschCQbyBF9SgYBaNgFIwoAACNV0JCf6X5wwAAAABJRU5ErkJggg==","orcid":"","institution":"The Affiliated Jiangsu Shengze Hospital of Nanjing Medical University","correspondingAuthor":true,"prefix":"","firstName":"Chenxu","middleName":"","lastName":"Xiao","suffix":""},{"id":594399826,"identity":"383226fe-9174-4b27-bdfc-cccc6e4f530e","order_by":1,"name":"Jing Shen","email":"","orcid":"https://orcid.org/0009-0001-5120-9897","institution":"The Affiliated Jiangsu Shengze Hospital of Nanjing Medical University","correspondingAuthor":false,"prefix":"","firstName":"Jing","middleName":"","lastName":"Shen","suffix":""},{"id":594399827,"identity":"a49042e0-93ab-4815-9c2c-32db0d0b92d7","order_by":2,"name":"Yuxuan Shen","email":"","orcid":"","institution":"","correspondingAuthor":false,"prefix":"","firstName":"Yuxuan","middleName":"","lastName":"Shen","suffix":""}],"badges":[],"createdAt":"2026-01-13 01:20:12","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-8585990/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-8585990/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":103437705,"identity":"708d4217-fdc2-43f3-a6a4-1bee87774c9d","added_by":"auto","created_at":"2026-02-25 16:51:28","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":1191992,"visible":true,"origin":"","legend":"\u003cp\u003eStudy design and analytic workflow.\u003c/p\u003e","description":"","filename":"Fig.1.png","url":"https://assets-eu.researchsquare.com/files/rs-8585990/v1/6322b2358d030fe6cb7f8f61.png"},{"id":103437706,"identity":"3f6a2b49-630e-4b74-a4d3-36b210ef3e0f","added_by":"auto","created_at":"2026-02-25 16:51:28","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":2784437,"visible":true,"origin":"","legend":"\u003cp\u003eGenetic sharing and polygenicity between Insomnia and ALPS. A. Univariate MiXeR estimates of heritability and polygenicity for each phenotype. B. multivariate MiXeR estimates the proportion of shared variants with concordant effect directions between Insomnia and ALPS. C. Proportion of shared effects between Insomnia and ALPS estimated by GPA. D. Venn diagram showing the number of shared variants between Insomnia and ALPS estimated by MiXeR. Counts of shared and phenotype-specific variants are shown in thousands, with standard errors indicated.\u003c/p\u003e","description":"","filename":"Fig.2.png","url":"https://assets-eu.researchsquare.com/files/rs-8585990/v1/209fd259108590fceac2bf29.png"},{"id":103507352,"identity":"bd5917db-bef8-4795-864b-754592d49beb","added_by":"auto","created_at":"2026-02-26 13:41:05","extension":"png","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":7505614,"visible":true,"origin":"","legend":"\u003cp\u003eLocal Genetic Correlations and Shared SNPs Between Insomnia and ALPS Phenotypes. A. Local genetic correlation between Insomnia and Left_ALPS and Mean_ALPS. Significant regions identified by LAVA analysis (after Benjamini-Hochberg correction) are indicated above the red dashed line. Black asterisks denote GWAS-PW significance (PPA3 \u0026gt; 0.95). B, C. Manhattan plots showing shared common genetic variants between Insomnia and Left_ALPS and between Insomnia and Mean_ALPS , based on the −log₁₀ transformed conjFDR values for each SNP. The solid horizontal line represents the reported shared association threshold (conjFDR \u0026lt; 0.05), and the dashed line indicates a more stringent threshold (conjFDR \u0026lt; 0.01). Lead SNPs are highlighted with red circles.\u003c/p\u003e","description":"","filename":"Fig.3.png","url":"https://assets-eu.researchsquare.com/files/rs-8585990/v1/845a397f55d1a5c711eedb97.png"},{"id":103437707,"identity":"ec3935fc-b6f2-4c15-a869-9fa71cbea935","added_by":"auto","created_at":"2026-02-25 16:51:28","extension":"png","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":7783315,"visible":true,"origin":"","legend":"\u003cp\u003eShared Genetic Variants Between Insomnia and ALPS and Their Gene Associations at 17q21.31. A. Manhattan plot showing shared common genetic variants between Insomnia and Right_ALPS, based on the −log₁₀ transformed conjFDR values for each SNP. The solid horizontal line indicates the reported shared association threshold (conjFDR \u0026lt; 0.05), while the dashed line represents a more stringent threshold (conjFDR \u0026lt; 0.01). Lead SNPs are highlighted with red circles. B. Sankey plot illustrating the relationships between SNPs at the 17q21.31 locus and their corresponding genes for each Insomnia and ALPS phenotype.\u003c/p\u003e","description":"","filename":"Fig.4.png","url":"https://assets-eu.researchsquare.com/files/rs-8585990/v1/c9418159cc3eda3ddd157b58.png"},{"id":103437710,"identity":"0ba7351f-d060-4870-a4ef-73ea934e0bf0","added_by":"auto","created_at":"2026-02-25 16:51:28","extension":"png","order_by":5,"title":"Figure 5","display":"","copyAsset":false,"role":"figure","size":19003044,"visible":true,"origin":"","legend":"\u003cp\u003eMulti-tissue and cell type-specific analyses of key genes associated with Insomnia and ALPS. A. Bubble plot showing multi-tissue TWAS results for key genes associated with Insomnia and ALPS. Positive Z_mean values indicate positive correlation, while negative Z_mean values indicate negative correlation. Significance is indicated as: FDR \u0026lt; 0.001, FDR \u0026lt; 0.01, FDR \u0026lt; 0.05. B. Multi-tissue SMR heatmap for \u003cem\u003eACBD4 \u003c/em\u003ewith Insomnia and ALPS. Significance is indicated as: FDR \u0026lt; 0.01, FDR \u0026lt; 0.05. C. Multi-tissue SMR bubble plot for \u003cem\u003eACBD4 \u003c/em\u003ewith Insomnia and ALPS. D. Cell type-specific SMR forest plot for \u003cem\u003eMAPT \u003c/em\u003ewith Insomnia and ALPS. E. Tissue enrichment heatmap based on PCGA analysis between Insomnia and ALPS. F. Bubble plot of gene-set enrichment analysis for key genes in specific brain cell types and pathological regions using the GSMAP method for Insomnia and ALPS.\u003c/p\u003e","description":"","filename":"Fig.5.png","url":"https://assets-eu.researchsquare.com/files/rs-8585990/v1/620b56145ac122c5195e804d.png"},{"id":103510337,"identity":"08af755a-427a-418a-b37c-718737d72a26","added_by":"auto","created_at":"2026-02-26 14:05:09","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":38160718,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-8585990/v1/0428f908-faf1-49ed-9e08-565e006bf6f1.pdf"},{"id":103437704,"identity":"8e6cc9ad-e148-48e7-b7fb-ba04f14c25fc","added_by":"auto","created_at":"2026-02-25 16:51:28","extension":"pdf","order_by":1,"title":"","display":"","copyAsset":false,"role":"supplement","size":346637,"visible":true,"origin":"","legend":"Supplementary Datas","description":"","filename":"SupplementaryDatas.pdf","url":"https://assets-eu.researchsquare.com/files/rs-8585990/v1/ee660032bd041fecdc220649.pdf"},{"id":103508120,"identity":"b2154a70-087c-4441-8fd3-0d3cb864dc77","added_by":"auto","created_at":"2026-02-26 13:47:16","extension":"docx","order_by":2,"title":"","display":"","copyAsset":false,"role":"supplement","size":12101622,"visible":true,"origin":"","legend":"Supplementary Figures","description":"","filename":"SupplementaryFigures.docx","url":"https://assets-eu.researchsquare.com/files/rs-8585990/v1/ef31717968af63fff639e425.docx"}],"financialInterests":"There is \u003cb\u003eNO\u003c/b\u003e Competing Interest.","formattedTitle":"A multilayer genomic framework linking insomnia to glymphatic system function through pleiotropic mechanisms at 17q21.31","fulltext":[{"header":"INTRODUCTION","content":"\u003cp\u003eSleep is fundamental to brain homeostasis, metabolic maintenance, and cognitive integrity(\u003cspan additionalcitationids=\"CR2 CR3\" citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e). Insomnia, one of the most prevalent sleep disorders, affects nearly one-third of adults(\u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e5\u003c/span\u003e) and has been shown by large-scale genome-wide association studies (GWAS) to involve a highly polygenic architecture spanning diverse biological pathways(\u003cspan additionalcitationids=\"CR7 CR8\" citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e9\u003c/span\u003e). Converging experimental and imaging evidence demonstrates that sleep promotes cerebrospinal fluid (CSF)\u0026ndash;interstitial fluid exchange and facilitates the clearance of neurotoxic metabolites, whereas sleep loss diminishes CSF influx and impairs the removal of proteins such as amyloid-β (Aβ) and phosphorylated tau (pTau)(\u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e, \u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e). These observations strongly suggest a mechanistic interface between sleep regulation and brain waste clearance.\u003c/p\u003e \u003cp\u003eThe glymphatic system (GS) constitutes a perivascular, glia-dependent transport network that drives the clearance of metabolic by-products, including Aβ and pTau(\u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e12\u003c/span\u003e, \u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e13\u003c/span\u003e). Despite strong physiological evidence for sleep\u0026ndash;GS coupling, it remains unknown whether insomnia and GS function share underlying genetic determinants, and whether any such overlap exhibits anatomical or molecular specificity. One major reason is that large-scale GWAS capable of resolving the genetic architecture of GS function have long been unavailable(\u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e14\u003c/span\u003e, \u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e15\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eThis barrier was overcome only recently with two population-level GWAS of MRI-derived diffusion tensor imaging along-the-perivascular-space (DTI-ALPS) phenotypes\u0026mdash;one characterizing hemispheric ALPS indices in more than 31,000 individuals(\u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e14\u003c/span\u003e), and another defining anatomically specific ALPS measures in more than 40,000 individuals(\u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e15\u003c/span\u003e). ALPS serves as a noninvasive imaging proxy of GS function by quantifying water diffusivity along perivascular white-matter tracts, thereby reflecting interstitial fluid transport mediated by astrocytic endfeet and perivascular pathways\u0026mdash;the core anatomical substrates of the GS function(\u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e14\u003c/span\u003e, \u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e15\u003c/span\u003e). These datasets therefore provide, for the first time, genetically informative and scalable proxies of GS function, enabling systematic evaluation of whether insomnia and GS converge at the genomic level and whether such convergence is spatially or biologically selective.\u003c/p\u003e \u003cp\u003eIn addition, prior studies suggest that GS activity primarily involves astrocytes and other neurovascular cell populations(\u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e12\u003c/span\u003e, \u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e16\u003c/span\u003e), raising the question of whether insomnia\u0026ndash;GS shared genetic signals preferentially map onto specific brain tissues or neural cell types. Finally, because GS function plays a central role in clearing Aβ and pTau(\u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e12\u003c/span\u003e, \u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e13\u003c/span\u003e), it is critical to determine whether genetic convergence between insomnia and GS aligns with regions or pathways vulnerable to neurodegenerative protein aggregation.\u003c/p\u003e"},{"header":"METHODS","content":"\u003cdiv id=\"Sec3\" class=\"Section2\"\u003e \u003ch2\u003eData Sources\u003c/h2\u003e \u003cdiv id=\"Sec4\" class=\"Section3\"\u003e \u003ch2\u003eInsomnia GWAS Dataset\u003c/h2\u003e \u003cp\u003eThe summary statistics for insomnia were obtained from a large-scale GWAS conducted by Lane et al. in the UK Biobank, with external validation in independent cohorts(\u003cspan citationid=\"CR17\" class=\"CitationRef\"\u003e17\u003c/span\u003e). The primary analysis was based on 453,379 individuals of European ancestry who self-reported their sleep patterns.\u003c/p\u003e \u003cp\u003eFor the GWAS, insomnia symptoms were defined using the question \u0026ldquo;Do you have trouble falling asleep at night, or do you wake up in the middle of the night?\u0026rdquo;. Participants were categorized into two primary phenotypes: those reporting \u0026ldquo;usually\u0026rdquo; (frequent insomnia) and those reporting \u0026ldquo;sometimes\u0026rdquo; or \u0026ldquo;usually\u0026rdquo; (any insomnia). This genetic association was robustly replicated in independent samples, including self-reported insomnia cases from the HUNT study and physician-diagnosed insomnia cases from the Partners Biobank, as well as in objective measures of sleep efficiency and duration from accelerometer data within the UK Biobank.\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e\n\u003ch3\u003eGS Function GWAS Datasets\u003c/h3\u003e\n\u003cp\u003eWe leveraged two recently published, comprehensive GWAS datasets on GS function as measured by the DTI-ALPS index.\u003c/p\u003e \u003cp\u003e \u003col\u003e \u003cspan\u003e \u003cli\u003e \u003cp\u003eFor the analysis of hemispheric ALPS indices, we used summary statistics from Huang et al.(\u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e14\u003c/span\u003e). This study analyzed data from 31,021 participants of white British ancestry in the UK Biobank, calculating separate indices for the left and right cerebral hemispheres (Left_ALPS, Right_ALPS) and their average (Mean_ALPS).\u003c/p\u003e \u003c/li\u003e \u003c/span\u003e \u003cspan\u003e \u003cli\u003e \u003cp\u003eFor the analysis of regional ALPS indices, we utilized data from Ran et al.(\u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e15\u003c/span\u003e). This research employed DTI-ALPS data from 40,486 European-ancestry individuals in the UK Biobank to define four distinct phenotypes based on anatomical location: anterior (aALPS), middle (mALPS), posterior (pALPS), and total (tALPS) ALPS indices.\u003c/p\u003e \u003c/li\u003e \u003c/span\u003e \u003c/ol\u003e \u003c/p\u003e\n\u003ch3\u003eQuality Control\u003c/h3\u003e\n\u003cp\u003ePrior to our analysis, the publicly available summary statistics underwent stringent quality control procedures. For the UK Biobank data, this included excluding individuals of non-European ancestry and those with incomplete genotype or phenotype data. At the variant level, single nucleotide polymorphisms (SNPs) were filtered based on minor allele frequency (MAF\u0026thinsp;\u0026lt;\u0026thinsp;0.01), Hardy-Weinberg equilibrium (\u003cem\u003eP\u003c/em\u003e\u0026thinsp;\u0026lt;\u0026thinsp;1 \u0026times; 10\u003csup\u003e\u0026minus;\u0026thinsp;6\u003c/sup\u003e), and imputation quality score (INFO\u0026thinsp;\u0026lt;\u0026thinsp;0.8). All genomic coordinates were harmonized to the GRCh37/hg19 reference assembly. The resulting dataset showed no significant inflation due to population stratification (LDSC intercept\u0026thinsp;\u0026asymp;\u0026thinsp;1.005), confirming its suitability for downstream integrative analyses.\u003c/p\u003e \u003cp\u003eThe full inclusion and exclusion criteria are provided in Supplementary Table\u0026nbsp;1, and an overview of the study design and analytic workflow is presented in Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003e.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003eInsert Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003e\u003c/p\u003e\n\u003ch3\u003eSpatial Transcriptomics (ST) Data for gsMap Analysis\u003c/h3\u003e\n\u003cp\u003eTo spatially contextualize the shared genetic signatures between insomnia and GS function, and to evaluate their relevance to the neuropathological continuum, we employed two complementary ST datasets.\u003c/p\u003e \u003cp\u003e \u003col\u003e \u003cspan\u003e \u003cli\u003e \u003cp\u003eHuman AD Pathology Dataset for Pathological Contextualization\u003c/p\u003e \u003c/li\u003e \u003c/span\u003e \u003c/ol\u003e \u003c/p\u003e \u003cp\u003eTo test the hypothesis that insomnia may influence AD risk through impairing GS-mediated clearance of pathological proteins, we leveraged ST data from postmortem AD brains(\u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e18\u003c/span\u003e). This dataset encompasses tissue sections from the inferior temporal cortex (ITC) of donors with advanced AD pathology (Braak V-VI) and normal controls, profiled using the 10x Genomics Visium platform. The key rationale for using this AD dataset is that it allows us to map the insomnia-GS shared genes directly onto the spatial epicenters of Aβ and pTau pathology. A significant enrichment in these pathological microenvironments would provide compelling spatial evidence linking our discovered genetic overlap to the core proteinopathic processes of AD.\u003c/p\u003e \u003cp\u003e\u003cspan\u003e\u0026nbsp;\u003c/span\u003e2. Mouse Embryonic Development Dataset for Investigating Developmental Origins\u003c/p\u003e \u003cp\u003eTo explore whether the insomnia-GS genetic overlap is embedded within fundamental neurodevelopmental programs, we analyzed data from the Mouse Organogenesis Spatiotemporal Transcriptomic Atlas (MOSTA)(\u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e19\u003c/span\u003e). We specifically focused on the embryonic day 16.5 (E16.5) sample (E16.5_E1S1.MOSTA.h5ad). The E16.5 stage represents a peak period of murine neurogenesis, gliogenesis, and overall brain architecture establishment. By examining the enrichment of insomnia-GS genes at this developmental critical window, we aimed to determine if their shared biological influence originates during early brain construction, potentially setting the stage for lifelong brain function and vulnerability to sleep and clearance disorders.\u003c/p\u003e \u003cdiv id=\"Sec8\" class=\"Section2\"\u003e \u003ch2\u003eBivariate causal mixture model (MiXeR) analysis\u003c/h2\u003e \u003cp\u003eWe used MiXeR (v2.2.1) to quantify the polygenic architecture of insomnia and each ALPS phenotype and to estimate the extent of their shared causal variant sets(\u003cspan additionalcitationids=\"CR21\" citationid=\"CR20\" class=\"CitationRef\"\u003e20\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e22\u003c/span\u003e). MiXeR applies a Gaussian mixture model to GWAS summary statistics to infer key parameters of genetic architecture. In MiXeR, π denotes the proportion of SNPs with non-zero causal effects (polygenicity), and σ\u0026sup2;β represents the variance of causal SNP effect sizes. For bivariate modeling, MiXeR additionally estimates π₁₂, the proportion of shared causal SNPs between two traits, and ρ_beta, the correlation of effect sizes among these shared causal variants. From these quantities, MiXeR also derives a model-based estimate of the genetic correlation (rg_MiXeR), which reflects the genetic correlation attributable specifically to shared causal components.\u003c/p\u003e \u003cp\u003eSummary statistics for insomnia and all seven ALPS phenotypes were subjected to unified quality control, and linkage disequilibrium (LD) was estimated using the 1000 Genomes Phase 3 European reference panel. Before model fitting, summary statistics were divided by chromosome using the split_sumstats function.\u003c/p\u003e \u003cp\u003eFor each trait, we first performed univariate model fitting using mixer.py fit1 to obtain trait-specific architecture parameters (π and σ\u0026sup2;β), which were subsequently used as priors in bivariate modeling with mixer.py fit2 to estimate shared versus trait-specific causal components (π₁₂ and ρ_beta). Model fit was evaluated relative to nested minimal-overlap (genetic-correlation\u0026ndash;only) and full-overlap models using AIC and BIC, ensuring that polygenic sharing was statistically justified rather than a modeling artifact. To enhance robustness, all analyses were repeated across 20 randomly sampled SNP subsets from the LD reference panel, and reported estimates represent averaged results. Final summaries, including Venn diagrams and Dice coefficients quantifying the degree of shared causal variants, were generated using the mixer_figures.py module.\u003c/p\u003e \u003c/div\u003e\n\u003ch3\u003eGenetic analysis integrating pleiotropy and functional annotation (GPA)\u003c/h3\u003e\n\u003cp\u003eWe applied GPA(\u003cspan citationid=\"CR23\" class=\"CitationRef\"\u003e23\u003c/span\u003e) to enhance the detection of shared risk variants across insomnia and each ALPS phenotype. GPA is a probabilistic mixture-model framework that jointly analyzes GWAS summary statistics from multiple related traits and integrates functional annotations to improve the prioritization of associated SNPs.\u003c/p\u003e \u003cp\u003eGPA models the distribution of GWAS p-values using a latent categorical mixture. For a single trait, p-values are assumed to arise from a mixture of a uniform distribution U(0, 1) for null SNPs and a Beta (α, 1) distribution (0\u0026thinsp;\u0026lt;\u0026thinsp;α\u0026thinsp;\u0026lt;\u0026thinsp;1) for non-null SNPs, reflecting the tendency of associated variants to generate small p-values. For two-trait analyses (K\u0026thinsp;=\u0026thinsp;2), SNPs are assigned to four latent association classes: non-associated with either trait (00), associated only with trait 1 (\u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e), associated only with trait 2 (01), or associated with both traits (\u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e). Model parameters were estimated via maximum likelihood.\u003c/p\u003e \u003cp\u003eFollowing model fitting, we computed local false discovery rates (FDR) for each SNP corresponding to its posterior probability of belonging to each association class. For each insomnia\u0026ndash;ALPS pair, SNPs were prioritized based on local FDR thresholds to identify variants associated with insomnia, with ALPS, or jointly with both traits while controlling the global FDR.\u003c/p\u003e \u003cp\u003eTo test whether significant pleiotropy existed between trait pairs, we performed a likelihood-ratio test comparing a null model assuming independence of association states\u0026mdash;i.e., π₁₁ = (π₁₀ + π₁₁) (π₀₁ + π₁₁) \u0026mdash; with an alternative model allowing non-independent enrichment of jointly associated variants. The test statistic, \u0026minus;\u0026thinsp;2 log λ(P), asymptotically follows a χ\u0026sup2; distribution with 1 degree of freedom. Significant deviation from the null supports true genetic pleiotropy between insomnia and the ALPS phenotype under study.\u003c/p\u003e\n\u003ch3\u003eLocal Analysis of [co]Variant Association (LAVA)\u003c/h3\u003e\n\u003cp\u003eWe also performed local genetic correlation analysis using LAVA(\u003cspan citationid=\"CR24\" class=\"CitationRef\"\u003e24\u003c/span\u003e). The genome was partitioned into 2,495 approximately independent LD blocks based on the 1000 Genomes European reference panel. For each block, LAVA estimates the local genetic correlation (ρ), defined as the correlation between the genetic effect sizes of two traits within a given LD block. Local ρ values were obtained using LAVA\u0026rsquo;s bivariate model after verifying sufficient univariate signal in both traits. To account for multiple testing, we applied a Bonferroni-corrected significance threshold of \u003cem\u003eP\u003c/em\u003e\u0026thinsp;\u0026lt;\u0026thinsp;0.05/2495\u0026thinsp;=\u0026thinsp;2.00 \u0026times; 10\u003csup\u003e\u0026minus;\u0026thinsp;5\u003c/sup\u003e.\u003c/p\u003e \u003cp\u003eIn addition, we applied the Pairwise GWAS (GWAS-PW) method to further assess whether genomic regions were shared between ALPS and insomnia(\u003cspan citationid=\"CR25\" class=\"CitationRef\"\u003e25\u003c/span\u003e). This approach estimates the posterior probability of a locus being associated with both traits (PPA3: shared causal variant; PPA4: distinct causal variants within the same locus). We considered a region as significantly shared if the posterior probability of colocalization (PPA3) exceeded 0.8, in accordance with the original GWAS-PW recommendations.\u003c/p\u003e \u003cdiv id=\"Sec11\" class=\"Section2\"\u003e \u003ch2\u003eCondFDR and conjFDR\u003c/h2\u003e \u003cp\u003eTo identify genetic loci jointly associated with insomnia and ALPS traits, we applied the condFDR and conjFDR framework(\u003cspan citationid=\"CR26\" class=\"CitationRef\"\u003e26\u003c/span\u003e, \u003cspan citationid=\"CR27\" class=\"CitationRef\"\u003e27\u003c/span\u003e). This empirical Bayes approach enhances discovery power by leveraging genetic pleiotropy between traits, re-ranking SNP association p-values from a primary GWAS based on enrichment patterns observed in a secondary GWAS.\u003c/p\u003e \u003cp\u003eWe computed the condFDR for insomnia and ALPS traits, and reciprocally for ALPS traits given insomnia, using publicly available MATLAB scripts. The condFDR for a given SNP is defined as the posterior probability that the SNP is null for the primary trait, given that its p-values in both traits are as small as or smaller than the observed values. The conjFDR, defined as the maximum of the two reciprocal condFDR values, was used to identify loci significantly associated with both traits simultaneously.\u003c/p\u003e \u003cp\u003eTo minimize bias from long-range LD, we excluded the extended major histocompatibility complex (MHC) region (chromosome 6: 25\u0026ndash;34 Mb) and the chromosomal region 8p23.1 (chromosome 8: 7.2\u0026ndash;12.5 Mb). All test statistics were corrected for genomic inflation using intergenic SNPs, which are relatively depleted of true associations. SNPs were randomly pruned over 500 iterations using an LD threshold of r\u003csup\u003e2\u003c/sup\u003e\u0026thinsp;\u0026gt;\u0026thinsp;0.1 to approximate independence and mitigate p-value inflation. A significance threshold of Benjamini\u0026ndash;Hochberg FDR\u0026thinsp;\u0026lt;\u0026thinsp;0.05 was applied to declare statistical significance. Independent genomic loci were defined using FUMA, based on LD blocks (r\u003csup\u003e2\u003c/sup\u003e\u0026thinsp;\u0026lt;\u0026thinsp;0.6) and merged within 250 kb windows.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec12\" class=\"Section2\"\u003e \u003ch2\u003eColocalization and fine-mapping analyses\u003c/h2\u003e \u003cp\u003eTo identify shared causal variants between insomnia and ALPS phenotypes, we performed a series of Bayesian and likelihood-based colocalization and fine-mapping analyses. First, we applied the coloc R package (v5)(\u003cspan citationid=\"CR28\" class=\"CitationRef\"\u003e28\u003c/span\u003e) to estimate the posterior probability that two traits share the same causal variant within each genomic region. Coloc evaluates five mutually exclusive models\u0026mdash;no association with either trait (H0), association with only one trait (H1/H2), association with both traits but via distinct causal variants (H3), or association with both traits via a shared causal variant (H4). Evidence for colocalization was defined by a high posterior probability for H4 (PP.H4), typically PP.H4\u0026thinsp;\u0026gt;\u0026thinsp;0.5 or \u0026gt;\u0026thinsp;0.8 depending on stringency. Analyses were performed using GWAS summary statistics together with LD matrices derived from the 1000 Genomes European reference panel.\u003c/p\u003e \u003cp\u003eBecause single-causal-variant assumptions may be violated in regions with complex LD or multiple independent signals, we additionally used the Sum of Single Effects (SuSiE) regression framework(\u003cspan citationid=\"CR29\" class=\"CitationRef\"\u003e29\u003c/span\u003e) implemented through coloc.susie. SuSiE decomposes trait associations into a sparse set of \u0026ldquo;single-effect\u0026rdquo; components, each corresponding to a putative causal SNP, and returns credible sets representing distinct association signals. We ran SuSiE separately for each trait using runsusie, and then used coloc.susie to assess whether credible sets from insomnia and ALPS overlapped. This approach is well suited for loci containing multiple independent causal variants.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec13\" class=\"Section2\"\u003e \u003ch2\u003eIntegration with FINEMAP and echolocatoR\u003c/h2\u003e \u003cp\u003eTo further fine-map causal configurations, we used FINEMAP (v1.4+)(\u003cspan citationid=\"CR30\" class=\"CitationRef\"\u003e30\u003c/span\u003e) as implemented within the echolocatoR pipeline(\u003cspan citationid=\"CR31\" class=\"CitationRef\"\u003e31\u003c/span\u003e). FINEMAP enumerates possible causal configurations (up to five causal variants per region by default) and computes posterior inclusion probabilities (PIP) for individual SNPs and posterior probabilities for multi-variant causal models. Colocalization was inferred when high-probability causal configurations for insomnia and ALPS shared one or more SNPs with large PIP values. All analyses used LD matrices derived from the European 1000 Genomes reference panel, and harmonized GWAS summary statistics including effect sizes, standard errors, allele frequencies, and genomic coordinates.\u003c/p\u003e \u003cp\u003eAcross methods, we prioritized colocalized signals supported by multiple approaches or yielding high-confidence metrics (e.g., PP.H4\u0026thinsp;\u0026gt;\u0026thinsp;0.8 for coloc, PIP\u0026thinsp;\u0026gt;\u0026thinsp;0.95 for SuSiE or FINEMAP). Such convergent multi-method evidence was used to define the set of credible shared causal variants at each locus.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec14\" class=\"Section2\"\u003e \u003ch2\u003eFine-mapping Of Causal eQTLs (FOCUS)\u003c/h2\u003e \u003cp\u003eTo identify putative causal genes underlying the shared genetic signals between insomnia and ALPS phenotypes, we applied FOCUS(\u003cspan citationid=\"CR32\" class=\"CitationRef\"\u003e32\u003c/span\u003e), a Bayesian framework that integrates GWAS summary statistics with tissue-specific eQTL reference panels. FOCUS models the joint distribution of GWAS and eQTL effect sizes to compute the PIP that each gene mediates the association signal within a given locus.\u003c/p\u003e \u003cp\u003eAnalyses were restricted to genomic regions showing evidence of shared genetic architecture based on conjFDR and LAVA results. For each region, we supplied FOCUS with harmonized GWAS summary statistics for insomnia and ALPS phenotypes, LD matrices derived from the 1000 Genomes Project Phase 3 European reference panel, and precomputed eQTL weight matrices from the focus.db repository, which aggregates GTEx-based expression prediction models across multiple tissues and cell types. Gene boundaries were defined using GENCODE v37 annotations(\u003cspan citationid=\"CR33\" class=\"CitationRef\"\u003e33\u003c/span\u003e, \u003cspan citationid=\"CR34\" class=\"CitationRef\"\u003e34\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eFOCUS returns a posterior distribution over genes that are compatible with the observed association pattern, quantifying for each gene the probability of being causal. Genes with PIP\u0026thinsp;\u0026gt;\u0026thinsp;0.8 were considered high-confidence causal candidates. These fine-mapping results were subsequently integrated with TWAS, Summary-based Mendelian Randomization (SMR), and colocalization findings to construct a convergent causal gene set for each shared insomnia\u0026ndash;ALPS locus.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec15\" class=\"Section2\"\u003e \u003ch2\u003eVariant annotation and functional characterization\u003c/h2\u003e \u003cp\u003eTo functionally characterize genetic variants associated with insomnia and ALPS phenotypes, we performed locus definition and variant-level annotation using the Functional Mapping and Annotation (FUMA) platform(\u003cspan citationid=\"CR35\" class=\"CitationRef\"\u003e35\u003c/span\u003e). Genomic risk loci were identified based on genome-wide significant SNPs (\u003cem\u003eP\u003c/em\u003e\u0026thinsp;\u0026lt;\u0026thinsp;5 \u0026times; 10⁻⁸) using LD information from the 1000 Genomes Project Phase 3 European reference panel. Independent significant SNPs were defined using FUMA\u0026rsquo;s LD-clumping procedure, and signals within 250 kb of each other were merged into a single locus to capture extended association blocks.\u003c/p\u003e \u003cp\u003eFunctional annotation of all SNPs mapped to each locus was conducted through FUMA\u0026rsquo;s ANNOVAR-based framework(\u003cspan citationid=\"CR36\" class=\"CitationRef\"\u003e36\u003c/span\u003e). We integrated multiple annotation resources to evaluate the potential functional impact of variants, including predicted deleteriousness scores and regulatory features. Variants with CADD scores\u0026thinsp;\u0026gt;\u0026thinsp;12.37 were considered to have potentially damaging effects, while RegulomeDB scores were used to infer regulatory activity, with lower scores indicating stronger evidence for regulatory function. Additional annotations, including chromatin states, eQTL mappings, and positional mappings to nearby genes, were incorporated to prioritize variants with plausible biological relevance to insomnia\u0026ndash;ALPS shared genetic architecture.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec16\" class=\"Section2\"\u003e \u003ch2\u003eTWAS analyses using S-PrediXcan and S-MultiXcan\u003c/h2\u003e \u003cp\u003eWe performed TWAS to identify genes whose genetically predicted expression is associated with insomnia and each ALPS phenotype, using S-PrediXcan and S-MultiXcan(\u003cspan citationid=\"CR37\" class=\"CitationRef\"\u003e37\u003c/span\u003e, \u003cspan citationid=\"CR38\" class=\"CitationRef\"\u003e38\u003c/span\u003e). S-PrediXcan was applied to estimate gene\u0026ndash;trait associations in whole blood using GTEx v8 pre-trained expression prediction models. This approach integrates GWAS summary statistics with cis-eQTL weights to infer whether genetically regulated gene expression contributes to trait variation.\u003c/p\u003e \u003cp\u003eTo improve power by leveraging shared regulatory effects across brain regions, we further applied S-MultiXcan, which jointly analyzes expression prediction models from 13 GTEx v8 brain tissues. These tissues included: Brain_Putamen_basal_ganglia, Brain_Frontal_Cortex_BA9, Brain_Hippocampus, Brain_Cerebellum, Brain_Anterior_cingulate_cortex_BA24, Brain_Caudate_basal_ganglia, Brain_Cerebellar_Hemisphere, Brain_Nucleus_accumbens_basal_ganglia, Brain_Spinal_cord_cervical_c-1, Brain_Amygdala, Brain_Substantia_nigra, Brain_Cortex, and Brain_Hypothalamus. Together, these GTEx v8 models encompass 838 donors and 15,201 RNA-seq samples(\u003cspan citationid=\"CR39\" class=\"CitationRef\"\u003e39\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eAll analyses used LD covariance matrices matched to the GTEx prediction models, derived from the 1000 Genomes Phase 3 European reference panel, and were conducted using the default settings of the PrediXcan/S-MultiXcan framework. Multiple testing correction was performed using the Benjamini\u0026ndash;Hochberg FDR, and genes with FDR-significant associations were retained as TWAS-prioritized candidates.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec17\" class=\"Section2\"\u003e \u003ch2\u003eSMR and Heterogeneity in Dependent Instruments (HEIDI) testing\u003c/h2\u003e \u003cp\u003eTo evaluate whether genetically regulated gene expression exerts causal effects on insomnia and ALPS phenotypes, we applied the SMR framework(\u003cspan citationid=\"CR40\" class=\"CitationRef\"\u003e40\u003c/span\u003e). SMR integrates GWAS and cis-eQTL summary statistics to estimate the effect of gene expression on complex traits under an instrumental-variable (IV) paradigm, using the top cis-eQTL of each gene as the IV. This approach enables causal inference without requiring individual-level genotype or transcriptomic data.\u003c/p\u003e \u003cp\u003eTo distinguish true causal relationships from spurious associations driven by LD, we performed the HEIDI test as implemented in the SMR software. Associations showing no evidence of heterogeneity (p_HEIDI\u0026thinsp;\u0026gt;\u0026thinsp;0.01) were interpreted as consistent with a shared causal variant, whereas signals failing the HEIDI test were excluded as potential LD-induced artifacts.\u003c/p\u003e \u003cp\u003eWe conducted SMR analyses using cis-eQTL panels from 14 GTEx v8 brain tissues and whole blood(\u003cspan citationid=\"CR39\" class=\"CitationRef\"\u003e39\u003c/span\u003e), enabling systematic assessment of tissue-specific expression\u0026ndash;trait associations across multiple neuroanatomical regions. To further evaluate the contribution of cell type\u0026ndash;specific regulatory mechanisms, we additionally incorporated cis-eQTL weights from nine neural cell populations reported by Bryois et al.(\u003cspan citationid=\"CR41\" class=\"CitationRef\"\u003e41\u003c/span\u003e). These datasets included excitatory and inhibitory neurons, astrocytes, oligodendrocytes, microglia, OPCs, endothelial cells, and other key neurovascular cell types.\u003c/p\u003e \u003cp\u003eAll SMR and HEIDI analyses were performed using default settings. Multiple testing correction was applied using the Benjamini\u0026ndash;Hochberg FDR, and genes passing both SMR significance and HEIDI filtering were retained as putative causal candidates for insomnia and ALPS phenotypes.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec18\" class=\"Section2\"\u003e \u003ch2\u003eMulti-Trait Analysis of GWAS (MTAG)\u003c/h2\u003e \u003cp\u003eTo increase statistical power and improve the discovery of shared and trait-specific association signals, we applied MTAG, an extension of univariate GWAS that jointly analyzes summary statistics from genetically correlated traits(\u003cspan citationid=\"CR42\" class=\"CitationRef\"\u003e42\u003c/span\u003e). MTAG models the vector of GWAS Z-scores across traits under a multivariate normal framework that accommodates genetic correlation and explicitly corrects for sample overlap, thereby enabling each trait to \u0026ldquo;borrow strength\u0026rdquo; from others to enhance effective association power.\u003c/p\u003e \u003cp\u003eWe followed the standard MTAG workflow, using harmonized GWAS summary statistics for insomnia and the seven ALPS phenotypes as input. MTAG was run using default parameter settings, and standard error corrections were applied to account for overlapping samples across input GWAS datasets. This procedure yielded MTAG-enhanced GWAS summary statistics for each ALPS phenotype conditioned on shared genetic information with insomnia.\u003c/p\u003e \u003cp\u003eAnalyses were performed in Python using the publicly available MTAG software released by Turley et al.(\u003cspan citationid=\"CR42\" class=\"CitationRef\"\u003e42\u003c/span\u003e). All MTAG outputs underwent standard quality checks and were subsequently used in downstream gene-mapping, tissue- and cell-type enrichment, and transcriptomic integration analyses.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec19\" class=\"Section2\"\u003e \u003ch2\u003eTissue- and cell-type enrichment analysis using Phenotype\u0026ndash;Cell\u0026ndash;Gene Association (PCGA)\u003c/h2\u003e \u003cp\u003eTo systematically evaluate the contribution of tissues and cell types to insomnia and GS function, we applied the PCGA framework(\u003cspan citationid=\"CR43\" class=\"CitationRef\"\u003e43\u003c/span\u003e). PCGA builds on the Driver tissue and gene Estimation (DESE) methodology to jointly infer trait-relevant tissues, cell types, and putative susceptibility genes by integrating GWAS summary statistics with large-scale bulk and single-cell transcriptomic resources. The PCGA reference compendium includes expression profiles from 54 human tissues, 2,214 human cell types, and 4,384 mouse cell types, allowing high-resolution inference across multiple biological hierarchies.\u003c/p\u003e \u003cp\u003eIn this study, we first generated MTAG-enhanced GWAS summary statistics for insomnia and each of the seven ALPS phenotypes. These MTAG-META results were then used as input to PCGA, which models GWAS association patterns jointly with tissue- and cell-type\u0026ndash;specific expression signatures to identify enriched biological compartments. Through its hierarchical DESE-based inference, PCGA estimates the tissues, cell types, and gene sets most likely to underlie trait-associated genetic signals.\u003c/p\u003e \u003cp\u003eAll analyses were performed using the default PCGA workflow, reference datasets, and significance evaluation procedures. Enriched tissues, cell populations, and inferred susceptibility genes were retained as candidates contributing to shared insomnia\u0026ndash;ALPS biology.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec20\" class=\"Section2\"\u003e \u003ch2\u003eSpatially resolved genetic mapping using gsMap\u003c/h2\u003e \u003cp\u003eWe applied genetically informed gsMap to integrate GWAS summary statistics with ST data and identify spatially localized cell populations and genes associated with insomnia, AD, and the seven ALPS phenotypes(\u003cspan citationid=\"CR44\" class=\"CitationRef\"\u003e44\u003c/span\u003e). gsMap first uses a graph neural network (GNN) to learn low-dimensional embeddings for each ST spot by jointly modeling gene expression profiles and spatial coordinates. Spots are then grouped into homogeneous microdomains based on cosine similarity, and gene-specificity scores (GSS)\u0026mdash;quantifying the spatial specificity of each gene\u0026mdash;are computed.\u003c/p\u003e \u003cp\u003eGSS values are mapped to nearby SNPs (\u0026plusmn;\u0026thinsp;50 kb around the transcription start site) or to distal variants using external chromatin-based enhancer annotations (e.g., ABC model). For each ST spot, gsMap applies stratified LD score regression (S-LDSC) to test whether SNPs with high GSS values are enriched for trait heritability. To evaluate regional associations (e.g., anatomical zones or microdomains), gsMap aggregates spot-level p-values using the Cauchy combination test to generate p_cauchy statistics. Additional metrics\u0026mdash;including p_median, the regional median GSS, and the Pearson correlation coefficient (PCC) between GSS and local expression\u0026mdash;provide complementary measures of spatial association strength and specificity.\u003c/p\u003e \u003cp\u003eIn this study, we analyzed GWAS summary statistics for insomnia, AD, and all seven ALPS phenotypes using two complementary ST datasets: (i) human postmortem ITC Visium data (Visium_SPG_AD) and (ii) E16.5 mouse brain MOSTA sections, reflecting early neuroglial and neurovascular development. Both spot-level and region-level analyses were performed to identify spatial hotspots, trait-associated microdomains, and candidate genes with anatomically constrained effects.\u003c/p\u003e \u003c/div\u003e"},{"header":"RESULTS","content":"\u003cdiv id=\"Sec22\" class=\"Section2\"\u003e \u003ch2\u003ePolygenic analyses reveal non-random genetic overlap between insomnia and ALPS phenotypes\u003c/h2\u003e \u003cp\u003eTo assess the genetic relationship between insomnia and ALPS, we first constructed cross-trait conditional Q\u0026ndash;Q plots. Conditioning on ALPS, insomnia displayed a marked upward deviation from the null among SNPs with increasing ALPS significance (e.g., p_ALPS\u0026thinsp;\u0026lt;\u0026thinsp;10⁻\u0026sup3;), and the reciprocal pattern was likewise observed. This bidirectional deviation indicates substantial polygenic enrichment between the two traits beyond chance expectations (Supplementary Fig.\u0026nbsp;1\u0026ndash;8).\u003c/p\u003e \u003cp\u003ePolygenic modeling provided additional support for this architecture. Although insomnia showed lower SNP-based heritability (h\u0026sup2; = 0.080) than ALPS phenotypes (h\u0026sup2; = 0.135\u0026ndash;0.189) (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003eA), the two traits exhibited comparable proportions of non-zero-effect SNPs (π) and similar effect size variances (σ\u0026sup2;β). Significant cross-trait conditional enrichment persisted, suggesting the presence of overlapping sets of risk-influencing variants (Supplementary Table\u0026nbsp;2).\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003eTo determine whether this enrichment reflects genuine shared genetic architecture rather than correlated noise, we applied bivariate MiXeR and GPA. Genome-wide genetic correlations were near zero (rg_MiXeR\u0026thinsp;=\u0026thinsp;\u0026minus;\u0026thinsp;0.038 to 0.003), yet both frameworks indicated modest but detectable overlap in causal variants (MiXeR Dice coefficients: 17.64%\u0026ndash;42.37%; GPA π₁₁: 6.44%\u0026ndash;9.07%) (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003eB, C, D). Nearly half of the shared SNPs (45.7%\u0026ndash;49.7%) exhibited opposite effect directions, consistent with antagonistic pleiotropy that may obscure global genetic correlations despite localized sharing. Importantly, all joint models significantly outperformed independence models (χ\u0026sup2; \u0026gt; 6300, p\u0026thinsp;\u0026lt;\u0026thinsp;10⁻\u0026sup3;⁰⁰), demonstrating robust, non-random, region-specific genetic overlap between insomnia and ALPS phenotypes (Supplementary Tables\u0026nbsp;3\u0026ndash;4).\u003c/p\u003e \u003cp\u003eInsert Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003e\u003c/p\u003e \u003cp\u003e \u003cb\u003eLocal genetic correlation and colocalization analyses identify 17q21.31 as a core locus shared between insomnia and ALPS phenotypes\u003c/b\u003e \u003c/p\u003e \u003cp\u003eLocal genetic correlation analysis using LAVA identified several regions with significant local genetic correlation (FDR\u0026thinsp;\u0026lt;\u0026thinsp;0.05), among which the chr17:42,348,004\u0026ndash;43,460,500 interval (17q21.31) emerged as the most prominent. This region showed strong positive local correlations between insomnia and both Left_ALPS (ρ\u0026thinsp;=\u0026thinsp;0.547, FDR\u0026thinsp;=\u0026thinsp;0.008) and Mean_ALPS (ρ\u0026thinsp;=\u0026thinsp;0.564, FDR\u0026thinsp;=\u0026thinsp;0.010), indicating focal convergence of genetic effects despite negligible genome-wide correlation. Bayesian colocalization using GWAS-PW further supported shared architecture at this locus: the posterior probability that insomnia and ALPS share a single causal variant (Model 3) was extremely high (PPA₃ \u0026gt; 0.99), providing strong evidence for true colocalization (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003eA; Supplementary Table\u0026nbsp;5). These findings collectively implicate 17q21.31 as a central pleiotropic hotspot influencing both insomnia susceptibility and GS function.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003eInsert Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003e\u003c/p\u003e \u003cdiv id=\"Sec23\" class=\"Section3\"\u003e \u003ch2\u003eConjFDR and Bayesian colocalization identify two independent shared causal signals within 17q21.31\u003c/h2\u003e \u003cp\u003eUsing the conjFDR framework (FDR\u0026thinsp;\u0026lt;\u0026thinsp;0.05), we identified 16 pleiotropic variants jointly associated with insomnia and at least one ALPS phenotype, among which two mapped to the 17q21.31 locus. The first signal, rs71373536\u0026mdash;located 259 bp downstream of \u003cem\u003eACBD4\u003c/em\u003e\u0026mdash;showed significant cross-trait enrichment for both insomnia\u0026ndash;Left_ALPS (conjFDR\u0026thinsp;=\u0026thinsp;0.004) and insomnia\u0026ndash;Mean_ALPS (conjFDR\u0026thinsp;=\u0026thinsp;0.004), with concordant effect directions across traits (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003eB, C). The second signal, rs1044977, a missense variant within \u003cem\u003eHEXIM1\u003c/em\u003e (CADD\u0026thinsp;=\u0026thinsp;18.79), was specifically detected for insomnia\u0026ndash;Right_ALPS (conjFDR\u0026thinsp;=\u0026thinsp;0.020) (Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003eA; Supplementary Table\u0026nbsp;6).\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003eBayesian colocalization further validated these findings. Within \u0026plusmn;\u0026thinsp;100 kb windows, rs71373536 exhibited strong evidence for a shared causal variant between insomnia and ALPS phenotypes (PP.H4\u0026thinsp;=\u0026thinsp;0.929\u0026ndash;0.936), whereas rs1044977 showed similarly high colocalization support (PP.H4\u0026thinsp;=\u0026thinsp;0.861). Both signals reside within the chr17:43.2\u0026ndash;43.3 Mb interval, overlapping the region independently highlighted by GWAS-PW (Supplementary Table\u0026nbsp;6).\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv id=\"Sec24\" class=\"Section2\"\u003e \u003ch2\u003eFine-mapping identifies multiple high-confidence causal variants, revealing the complex LD structure of 17q21.31\u003c/h2\u003e \u003cp\u003eTo further evaluate the number and confidence of putative causal variants underlying the two pleiotropic signals identified by conjFDR (rs1044977 and rs71373536), we performed fine-mapping within \u0026plusmn;\u0026thinsp;100 kb of each lead SNP. In the rs1044977 region, both FINEMAP and SuSiE converged on a model supporting five independent causal variants. Among these, rs9894577 (chr17:43,223,292) was consistently assigned the highest posterior probability (PP\u0026thinsp;\u0026asymp;\u0026thinsp;1.0) and represents the strongest causal candidate, located near the \u003cem\u003eACBD4\u0026ndash;HEXIM1\u0026ndash;PLCD3\u003c/em\u003e gene cluster. Although rs1044977 itself was not included in the top credible set, it is in very strong LD with rs9894577 (r\u0026sup2; \u0026gt; 0.95) and carries a high CADD score, suggesting that it may act as a functional proxy or contribute directly to the causal signal.\u003c/p\u003e \u003cp\u003eFine-mapping in the rs71373536 region similarly supported five independent causal components. The variant rs6503419 (chr17:43,149,988) emerged as the dominant causal signal (PP\u0026thinsp;=\u0026thinsp;1.0), residing near the \u003cem\u003eDCAKD\u0026ndash;NMT1\u0026ndash;PLCD3\u003c/em\u003e interval. rs71373536 is strongly correlated with rs6503419 (r\u0026sup2; \u0026gt; 0.8), indicating that these two SNPs likely tag the same underlying causal configuration (Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003eB; Supplementary Table\u0026nbsp;7).\u003c/p\u003e \u003cp\u003eTogether, these results highlight that the 17q21.31 locus contains at least two tightly linked but potentially functionally distinct shared causal signals, consistent with the region\u0026rsquo;s well-recognized extended LD structure and complex haplotype architecture.\u003c/p\u003e \u003cp\u003eInsert Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003e\u003c/p\u003e \u003cdiv id=\"Sec25\" class=\"Section3\"\u003e \u003ch2\u003eGene prioritization and causal inference identify key shared genes within 17q21.31\u003c/h2\u003e \u003cp\u003eTo functionally resolve the shared genetic signals at 17q21.31, we performed gene-level fine-mapping using FOCUS. Two genes\u0026mdash;\u003cem\u003eHEXIM1\u003c/em\u003e and \u003cem\u003eEFTUD2\u003c/em\u003e\u0026mdash;emerged with high posterior inclusion probabilities across multiple brain tissues. \u003cem\u003eHEXIM1\u003c/em\u003e showed consistently strong support (PIP\u0026thinsp;\u0026gt;\u0026thinsp;0.94 across cerebellar hemisphere, anterior cingulate cortex, and whole-blood models) and exhibited a significant negative TWAS association with insomnia (Z = \u0026minus;\u0026thinsp;7.43), indicating that reduced expression increases disease risk. Its missense variant (rs1044977, CADD\u0026thinsp;=\u0026thinsp;18.79) further suggests functional perturbation, nominating \u003cem\u003eHEXIM1\u003c/em\u003e as a key mediator linking sleep disturbance with neurodegenerative pathways. \u003cem\u003eEFTUD2\u003c/em\u003e also showed strong evidence (PIP\u0026thinsp;\u0026gt;\u0026thinsp;0.95 across frontal cortex, hippocampus, and dorsolateral prefrontal cortex), with expression positively associated with insomnia risk (Z\u0026thinsp;=\u0026thinsp;7.92) (Supplementary Table\u0026nbsp;8).\u003c/p\u003e \u003cp\u003eTo complement fine-mapping, we applied S-PrediXcan focusing on genes annotated to colocalized lead SNPs within 17q21.31 (Supplementary Table\u0026nbsp;9). In whole blood, \u003cem\u003eACBD4\u003c/em\u003e expression was significantly associated with insomnia (Z\u0026thinsp;=\u0026thinsp;4.02, FDR\u0026thinsp;=\u0026thinsp;0.0003), and also with Right_ALPS (Z\u0026thinsp;=\u0026thinsp;3.19, FDR\u0026thinsp;=\u0026thinsp;0.007) and Mean_ALPS (Z\u0026thinsp;=\u0026thinsp;2.97, FDR\u0026thinsp;=\u0026thinsp;0.015) (Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003eA; Supplementary Table\u0026nbsp;10). SMR analysis confirmed this pattern: \u003cem\u003eACBD4\u003c/em\u003e (top SNP: rs2291447) showed significant positive causal effects on insomnia and multiple ALPS phenotypes (b_SMR\u0026thinsp;\u0026gt;\u0026thinsp;0, FDR\u0026thinsp;\u0026lt;\u0026thinsp;0.05), suggesting that \u003cem\u003eACBD4\u003c/em\u003e-mediated processes in peripheral blood may contribute to the shared genetic architecture between insomnia and GS function (Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003eB, C; Supplementary Table\u0026nbsp;11).\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003eUsing S-MultiXcan across 13 GTEx brain tissues, three genes\u0026mdash;\u003cem\u003eHEXIM1\u003c/em\u003e, \u003cem\u003eACBD4\u003c/em\u003e, and \u003cem\u003eMAPT\u003c/em\u003e\u0026mdash;displayed significant associations for insomnia and at least one ALPS phenotype (FDR\u0026thinsp;\u0026lt;\u0026thinsp;0.05).\u003c/p\u003e \u003cp\u003e \u003cem\u003eHEXIM1\u003c/em\u003e remained the most strongly supported gene, showing consistent negative associations with insomnia (Z_mean = \u0026minus;\u0026thinsp;6.77, FDR\u0026thinsp;=\u0026thinsp;1.56\u0026times;10⁻\u0026sup1;⁵) and with Left_ALPS, Mean_ALPS, and Right_ALPS (Z_mean = \u0026minus;\u0026thinsp;2.63 to \u0026minus;\u0026thinsp;2.81), with strongest signals arising in basal ganglia regions (Putamen and Caudate), in agreement with FOCUS results (Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003eA; Supplementary Table\u0026nbsp;12).\u003c/p\u003e \u003cp\u003e \u003cem\u003eACBD4\u003c/em\u003e showed a significant positive association with insomnia (Z_mean\u0026thinsp;=\u0026thinsp;3.06, FDR\u0026thinsp;=\u0026thinsp;6.43\u0026times;10⁻\u0026sup1;\u0026sup1;), with its strongest tissue signal localized to the nucleus accumbens (Brain_Nucleus_accumbens_basal_ganglia, p_i_best\u0026thinsp;=\u0026thinsp;6.05\u0026times;10\u003csup\u003e\u0026minus;\u0026thinsp;6\u003c/sup\u003e). \u003cem\u003eACBD4\u003c/em\u003e expression was also positively associated with Mean_ALPS and Right_ALPS (Z_mean\u0026thinsp;=\u0026thinsp;2.49\u0026ndash;2.60, FDR\u0026thinsp;=\u0026thinsp;0.045\u0026ndash;0.048), with peak associations observed in the hypothalamus (p_i_best\u0026thinsp;=\u0026thinsp;8.42\u0026times;10⁻⁴\u0026ndash;8.66\u0026times;10⁻⁴), consistent with S-PrediXcan results (Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003eA; Supplementary Table\u0026nbsp;12).\u003c/p\u003e \u003cp\u003e \u003cem\u003eMAPT\u003c/em\u003e exhibited a significant positive association with insomnia (Z_mean\u0026thinsp;=\u0026thinsp;1.63, FDR\u0026thinsp;=\u0026thinsp;0.0071), with the strongest signal observed in the cervical spinal cord. In contrast, its association with Left_ALPS was negative (Z_mean = \u0026minus;\u0026thinsp;1.04, FDR\u0026thinsp;=\u0026thinsp;0.016), with the peak association arising from the caudate nucleus (Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003eA; Supplementary Table\u0026nbsp;12).\u003c/p\u003e \u003cp\u003eSMR analyses across 13 brain tissues and nine neural cell types further delineated the tissue- and cell-specific architecture of the shared loci. \u003cem\u003eACBD4\u003c/em\u003e showed consistent positive causal effects on insomnia and multiple ALPS phenotypes across several central nervous system regions, including the cerebellum, cerebellar hemisphere, and cervical spinal cord (b_SMR\u0026thinsp;\u0026gt;\u0026thinsp;0, FDR\u0026thinsp;\u0026lt;\u0026thinsp;0.05). These findings align with the TWAS results and indicate that higher \u003cem\u003eACBD4\u003c/em\u003e expression in these regions is associated with increased insomnia risk as well as elevated Right_ALPS and Mean_ALPS metrics (Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003eB, C; Supplementary Table\u0026nbsp;13).\u003c/p\u003e \u003cp\u003eIn contrast, \u003cem\u003eMAPT\u003c/em\u003e expression in astrocytes displayed significant negative causal effects on aALPS, mALPS, Mean_ALPS, pALPS, Right_ALPS and tALPS (b_SMR\u0026thinsp;\u0026lt;\u0026thinsp;0, FDR\u0026thinsp;\u0026lt;\u0026thinsp;0.05). This pattern is consistent with the S-MultiXcan results and indicates that astrocytic \u003cem\u003eMAPT\u003c/em\u003e expression is associated with reduced glymphatic-related diffusivity across multiple ALPS measures (Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003eD; Supplementary Table\u0026nbsp;14).\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec26\" class=\"Section3\"\u003e \u003ch2\u003eSpatial, tissue, and cell-type enrichment analyses reveal anatomically distributed but phenotype-specific substrates of insomnia\u0026ndash;GS pleiotropy\u003c/h2\u003e \u003cp\u003eIntegrative genetic analyses identified \u003cem\u003eHEXIM1\u003c/em\u003e and \u003cem\u003eACBD4\u003c/em\u003e at the 17q21.31 locus as shared molecular hubs for insomnia and ALPS phenotypes, and further revealed significant associations with \u003cem\u003eMAPT\u003c/em\u003e\u0026mdash;the core pathogenic gene of AD. Because 17q21.31 represents one of the most intensively characterized AD risk regions and \u003cem\u003eMAPT\u003c/em\u003e-driven tau pathology is a hallmark of AD, the presence of convergent signals at this locus suggested that pleiotropic variants influencing insomnia\u0026ndash;ALPS coupling may also intersect with AD-related pathogenic processes. Given that ALPS phenotypes index GS function, including clearance of neurotoxic proteins such as Aβ and pTau, we sought to spatially contextualize these shared genetic signals within an AD-relevant microenvironment. To this end, we analyzed ST data from the ITC of AD donors using gsMap to map the spatial and cellular distribution of trait-associated genes.\u003c/p\u003e \u003cp\u003eApplying gsMap to the ITC dataset revealed that aALPS exhibited exceptionally strong enrichment across multiple neuronal and glial populations, with the most pronounced signals in excitatory neurons (p_cauchy\u0026thinsp;=\u0026thinsp;9.60\u0026times;10⁻⁶⁴) and inhibitory neurons (p_cauchy\u0026thinsp;=\u0026thinsp;6.59\u0026times;10⁻\u0026sup1;⁸), alongside robust enrichment in astrocytes, microglia, and OPCs (all p_cauchy\u0026thinsp;\u0026lt;\u0026thinsp;5\u0026times;10⁻⁷). Notably, aALPS-associated genetic signals localized not only to Aβ and pTau deposition cores but also to surrounding periplaque regions (n_Aβ and n_pTau, denoting regions \u0026ldquo;next to Aβ/pTau deposits\u0026rdquo; in the original ST annotation), suggesting that these variants may influence broader microenvironmental responses to protein aggregation rather than the deposition process itself (Supplementary Table\u0026nbsp;15).\u003c/p\u003e \u003cp\u003eOther ALPS phenotypes displayed more anatomically circumscribed patterns. Left_ALPS (p_cauchy\u0026thinsp;=\u0026thinsp;1.39\u0026times;10⁻\u0026sup1;⁴, p_median\u0026thinsp;=\u0026thinsp;0.84) and Mean_ALPS (p_cauchy\u0026thinsp;=\u0026thinsp;7.74\u0026times;10⁻⁶, p_median\u0026thinsp;=\u0026thinsp;0.84) were most strongly enriched within Aβ-positive regions, whereas tALPS showed peak enrichment in astrocytes (p_cauchy\u0026thinsp;=\u0026thinsp;8.94\u0026times;10⁻\u0026sup1;⁰, p_median\u0026thinsp;=\u0026thinsp;0.81). In contrast, insomnia exhibited no significant enrichment in any spatial domain or cell type, while AD GWAS signals were highly enriched in microglia and multiple Aβ/pTau pathological zones (all p_cauchy\u0026thinsp;\u0026lt;\u0026thinsp;5\u0026times;10⁻⁷), providing an important reference for interpreting ALPS\u0026ndash;AD convergence (Supplementary Table\u0026nbsp;15).\u003c/p\u003e \u003cp\u003eTo validate these observations and further delineate the shared genetic architecture, we performed a cross-trait MTAG meta-analysis combining insomnia with each ALPS phenotype, followed by tissue and cell-type enrichment using PCGA. Shared genes identified by MTAG showed widespread enrichment across diverse brain regions\u0026mdash;including hippocampus, BA9, caudate, putamen, nucleus accumbens, BA24, amygdala, hypothalamus, cerebellum, cervical spinal cord, and substantia nigra\u0026mdash;and across major neural lineages, including excitatory and inhibitory neurons, astrocytes, microglia, oligodendrocytes and OPCs, vascular\u0026ndash;pia populations, endothelial cells, and retinal ganglion cells (P_adj\u0026thinsp;\u0026lt;\u0026thinsp;5\u0026times;10⁻⁷). These results were highly consistent across insomnia\u0026ndash;aALPS, \u0026ndash;Left_ALPS, \u0026ndash;Mean_ALPS, and \u0026ndash;tALPS shared gene sets, supporting a distributed neuroanatomical and multi-lineage cellular architecture underlying insomnia\u0026ndash;GS pleiotropy (Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003eE; Supplementary Table\u0026nbsp;16\u0026ndash;17).\u003c/p\u003e \u003cp\u003eCross-species gsMap analysis using E16.5 mouse MOSTA sections further corroborated these spatial enrichment patterns. Shared insomnia\u0026ndash;ALPS genes were significantly enriched in the spinal cord, brain, and choroid plexus (p_cauchy\u0026thinsp;\u0026lt;\u0026thinsp;5\u0026times;10⁻⁷), key tissues involved in early neuroglial maturation and the anatomical substrates that later support GS function (Supplementary Table\u0026nbsp;18).\u003c/p\u003e \u003cp\u003eTo further resolve gene-level spatial signatures, we profiled three key genes. \u003cem\u003eAQP4\u003c/em\u003e, a canonical structural component of the GS pathway, showed strong enrichment in Ab/pTau-adjacent regions (enrichment score\u0026thinsp;=\u0026thinsp;1.88, FDR\u0026thinsp;\u0026lt;\u0026thinsp;0.01) and in astrocytes (score\u0026thinsp;=\u0026thinsp;2.56, FDR\u0026thinsp;\u0026lt;\u0026thinsp;0.01). Notably, \u003cem\u003eAQP4\u003c/em\u003e expression correlated positively with all ALPS indices (r\u0026thinsp;\u0026gt;\u0026thinsp;0, P_bon\u0026thinsp;\u0026lt;\u0026thinsp;0.05) except aALPS, for which the correlation was negative (r\u0026thinsp;\u0026lt;\u0026thinsp;0, P_bon\u0026thinsp;\u0026lt;\u0026thinsp;0.05). This opposite direction of association relative to other ALPS measures highlights a marked regional specificity in the physiological processes captured by different ALPS indices (Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003eF; Supplementary Table\u0026nbsp;19).\u003c/p\u003e \u003cp\u003e \u003cem\u003eMAPT\u003c/em\u003e was strongly enriched in excitatory neurons (enrichment score\u0026thinsp;=\u0026thinsp;1.94, FDR\u0026thinsp;\u0026lt;\u0026thinsp;0.01) as well as in regions adjacent to both Aβ and pTau pathology (score\u0026thinsp;=\u0026thinsp;1.84, FDR\u0026thinsp;\u0026lt;\u0026thinsp;0.01). Within excitatory neurons, \u003cem\u003eMAPT\u003c/em\u003e expression showed negative correlations with all ALPS phenotypes (r\u0026thinsp;\u0026lt;\u0026thinsp;0, P_bon\u0026thinsp;\u0026lt;\u0026thinsp;0.05) except aALPS, for which the correlation was positive (r\u0026thinsp;\u0026gt;\u0026thinsp;0, P_bon\u0026thinsp;\u0026lt;\u0026thinsp;0.05) (Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003eF; Supplementary Table\u0026nbsp;19).\u003c/p\u003e \u003cp\u003e \u003cem\u003eHEXIM1\u003c/em\u003e showed enrichment in excitatory neurons (score\u0026thinsp;=\u0026thinsp;1.74, FDR\u0026thinsp;\u0026lt;\u0026thinsp;0.01) and Aβ regions (score\u0026thinsp;=\u0026thinsp;1.74, FDR\u0026thinsp;\u0026lt;\u0026thinsp;0.01), and consistently negative correlations with all ALPS phenotypes (r\u0026thinsp;\u0026lt;\u0026thinsp;0, P_bon\u0026thinsp;\u0026lt;\u0026thinsp;0.05), fully concordant with TWAS and SMR results (Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003eF; Supplementary Table\u0026nbsp;19).\u003c/p\u003e \u003cp\u003eTogether, these spatial and cellular analyses demonstrate that insomnia\u0026ndash;ALPS shared genes\u0026mdash;particularly those at 17q21.31\u0026mdash;are distributed across multiple neuronal and glial compartments, yet exhibit sharply phenotype-dependent and region-specific associations with GS function. These findings highlight that genetic effects on GS are highly context-dependent, varying across spatial microenvironments, cell types, and AD-related pathological regions.\u003c/p\u003e \u003cp\u003eInsert Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003e\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e"},{"header":"DISCUSSION","content":"\u003cp\u003eBy integrating large-scale GWAS of insomnia with hemispheric and region-specific ALPS phenotypes, this study provides a systematic genetic framework linking sleep disturbance to GS function. We demonstrate that: (i) insomnia and ALPS exhibit significant polygenic enrichment despite negligible genome-wide genetic correlation; (ii) the extended LD block at 17q21.31 constitutes a major shared susceptibility locus; (iii) multiple genes within this region, including \u003cem\u003eHEXIM1\u003c/em\u003e, \u003cem\u003eACBD4\u003c/em\u003e, \u003cem\u003eEFTUD2\u003c/em\u003e, and \u003cem\u003eMAPT\u003c/em\u003e, show convergent associations across statistical layers; (iv) shared genetic signals derived from MTAG meta-analysis display broad yet structured enrichment across diverse brain regions and neuroglial cell populations; and (v) The astrocyte-specific eQTL effect of \u003cem\u003eMAPT\u003c/em\u003e, together with the known role of glia in glymphatic regulation, suggests a potential glial pathway through which insomnia may influence perivascular clearance. Together, these findings define a multilayer genetic link between insomnia traits and GS function, extending molecular hypotheses connecting sleep disruption, glial dysfunction, and vulnerability to neurodegenerative processes.\u003c/p\u003e \u003cdiv id=\"Sec28\" class=\"Section2\"\u003e \u003ch2\u003eTargeted rather than genome-wide genetic coupling between insomnia and GS\u003c/h2\u003e \u003cp\u003eAlthough insomnia and ALPS showed negligible genome-wide genetic correlation, conjFDR revealed robust polygenic overlap. This pattern indicates that the two traits do not share a broadly diffuse genetic architecture, but instead converge through a targeted subset of molecular pathways. Mixture-model results, showing a large fraction of variants with opposite effects across traits, further support the view that GS-related biology forms a distinct yet functionally relevant component of the broader insomnia genetic landscape. These findings help explain why prior GWAS studies of sleep and neurological traits detected limited overlap(\u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e, \u003cspan citationid=\"CR45\" class=\"CitationRef\"\u003e45\u003c/span\u003e): shared biology exists, but its genetic footprint is localized rather than global, necessitating enrichment-based analytic strategies.\u003c/p\u003e \u003cp\u003e \u003cb\u003e17q21.31 as a shared genetic hub linking insomnia, GS, and neurodegeneration\u003c/b\u003e \u003c/p\u003e \u003cp\u003eA central contribution of this work is the identification of 17q21.31 as the major locus jointly influencing insomnia and GS function. This region spans the \u003cem\u003eMAPT\u003c/em\u003e H1/H2 inversion haplotype, characterized by extended LD, dense regulatory elements, and broad pleiotropic effects in AD, Parkinson\u0026rsquo;s disease, and frontotemporal dementia(\u003cspan additionalcitationids=\"CR47\" citationid=\"CR46\" class=\"CitationRef\"\u003e46\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR48\" class=\"CitationRef\"\u003e48\u003c/span\u003e). Because of the inversion, statistical associations largely reflect haplotype-level architecture rather than resolvable independent signals. Within this constraint, fine-mapping, TWAS, and SMR provide convergent functional prioritization rather than definitive causal separation. Fine-mapping highlighted \u003cem\u003eHEXIM1\u003c/em\u003e and \u003cem\u003eEFTUD2\u003c/em\u003e as top posterior candidates, while TWAS and SMR pointed to \u003cem\u003eACBD4\u003c/em\u003e and \u003cem\u003eMAPT\u003c/em\u003e. These findings suggest that multiple regulatory elements embedded in the 17q21.31 haplotype may contribute to insomnia\u0026ndash;GS biology, even though the underlying genetic architecture remains haplotype-driven.\u003c/p\u003e \u003cp\u003eMechanistically, these prioritized genes converge on pathways involving transcriptional regulation (\u003cem\u003eHEXIM1\u003c/em\u003e)(\u003cspan citationid=\"CR49\" class=\"CitationRef\"\u003e49\u003c/span\u003e), peroxisomal and lipid metabolism (\u003cem\u003eACBD4\u003c/em\u003e)(\u003cspan citationid=\"CR50\" class=\"CitationRef\"\u003e50\u003c/span\u003e), RNA splicing (\u003cem\u003eEFTUD2\u003c/em\u003e)(\u003cspan citationid=\"CR51\" class=\"CitationRef\"\u003e51\u003c/span\u003e), and tau-related astroglial function (\u003cem\u003eMAPT\u003c/em\u003e). The astrocyte-specific negative causal effect of \u003cem\u003eMAPT\u003c/em\u003e on ALPS indices is particularly notable, highlighting a glial pathway through which insomnia may impair perivascular clearance. Prior ALPS GWAS examining neurodegenerative biomarkers have identified a shared \u003cem\u003eMAPT\u003c/em\u003e locus between CSF p-tau and ALPS measures (rs7521; condFDR\u0026thinsp;=\u0026thinsp;0.0071), indicating that tau-related processes and GS function converge at the same 17q21.31 haplotype(\u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e14\u003c/span\u003e). This aligns with experimental evidence that tau pathology disrupts astrocytic endfeet and aquaporin-4 polarity(\u003cspan citationid=\"CR52\" class=\"CitationRef\"\u003e52\u003c/span\u003e), thereby compromising perivascular solute transport.\u003c/p\u003e \u003cp\u003eImportantly, external evidence from prior ALPS GWAS further reinforces the role of 17q21.31 in sleep\u0026ndash;GS coupling(\u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e14\u003c/span\u003e). In the ALPS dataset, independent significant variants within the \u003cem\u003eMAPT\u003c/em\u003e inversion block (rs62061734 and rs1991556) were previously reported as genome-wide significant loci for sleep duration in two large-scale GWAS (N\u0026thinsp;\u0026asymp;\u0026thinsp;1.3\u0026nbsp;million) conducted by Jansen et al.(\u003cspan citationid=\"CR53\" class=\"CitationRef\"\u003e53\u003c/span\u003e) and Doherty et al.(\u003cspan citationid=\"CR54\" class=\"CitationRef\"\u003e54\u003c/span\u003e) These SNPs map to the same region implicated in our insomnia\u0026ndash;GS analyses, indicating that sleep duration, insomnia liability, and GS-related perivascular clearance share convergent regulatory architecture within the 17q21.31 haplotype. Together, these findings position 17q21.31 as a biological nexus integrating sleep regulation, GS function, and vulnerability to neurodegenerative processes.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec29\" class=\"Section2\"\u003e \u003ch2\u003eDistributed but directionally divergent architecture across brain regions and cell types\u003c/h2\u003e \u003cp\u003eA notable aspect of our findings is that the shared genetic architecture between insomnia and GS function does not manifest as regional specificity. PCGA analyses revealed that shared genes were broadly enriched across multiple brain regions\u0026mdash;including the hippocampus, prefrontal cortex, basal ganglia, hypothalamus, cerebellum, and spinal cord\u0026mdash;and across diverse neuronal and glial lineages. This pattern indicates a distributed neuroanatomical and cellular footprint, rather than enrichment restricted to any single GS-relevant compartment.\u003c/p\u003e \u003cp\u003eHowever, fine-grained analyses revealed a different organizational principle. TWAS, SMR, and gsMap indicated that the direction of genetic effects frequently diverged across ALPS phenotypes and anatomical contexts, even when involving the same gene. \u003cem\u003eMAPT\u003c/em\u003e provides a clear example: although consistently showing negative associations with most ALPS indices across methods, its effect became positive for aALPS specifically within excitatory neurons. This illustrates that a single GS-related gene can exert opposite influences depending on local neurovascular and cellular environments. \u003cem\u003eAQP4\u003c/em\u003e, a canonical astrocytic water-channel gene(\u003cspan citationid=\"CR55\" class=\"CitationRef\"\u003e55\u003c/span\u003e), showed a similar pattern in our gsMap results\u0026mdash;positive associations with most hemispheric and posterior/middle ALPS indices but a negative association with aALPS. Although \u003cem\u003eAQP4\u003c/em\u003e is not among the core shared loci between insomnia and GS function, its well-established role in perivascular flux supports the biological plausibility of such context-dependent directional reversals.\u003c/p\u003e \u003cp\u003eTogether, these results support a distributed-but-directionally-divergent model of insomnia\u0026ndash;GS genetic coupling. Shared genes are broadly expressed across the brain, but their functional effects on perivascular clearance are shaped by local neurovascular and cellular context. This framework aligns with the compartmentalized nature of GS architecture(\u003cspan citationid=\"CR56\" class=\"CitationRef\"\u003e56\u003c/span\u003e), where regional differences in perivascular spacing, arterial pulsatility, and astrocytic polarity can invert genetic effects even when the same molecular pathways are involved. Thus, instead of classical spatial specificity, the insomnia\u0026ndash;GS interface is characterized by context-dependent functional heterogeneity across the brain.\u003c/p\u003e \u003c/div\u003e\n\u003ch3\u003eCross-species ST mapping highlights developmental neuroglial substrates\u003c/h3\u003e\n\u003cp\u003eTo refine the anatomical context of shared pathways, we leveraged mouse gsMap using MTAG-derived shared genes. The use of E16.5 mouse ST data is biologically justified because this developmental stage represents a critical window during which astrocytes, radial glia, and neurovascular scaffolding are actively established\u0026mdash;cellular and structural precursors that later form the core architecture of glymphatic pathways(\u003cspan citationid=\"CR57\" class=\"CitationRef\"\u003e57\u003c/span\u003e). Importantly, gene-expression gradients at this stage are highly conserved across species and often predict adult regional specialization.\u003c/p\u003e \u003cp\u003eShared insomnia\u0026ndash;ALPS genes showed selective localization within cortical and subcortical territories characterized by high metabolic demand and dense neuroglial interactions, mirroring the adult brain regions implicated in our PCGA and gsMap analyses. This concordance suggests that the molecular interface between sleep regulation and GS function may be rooted in early neurodevelopmental programs that specify astroglial polarity, perivascular organization, and neurovascular coupling. In this framework, embryonic spatial patterning provides the foundation upon which adult GS efficiency and sleep-related clearance vulnerability are built, offering a developmental perspective on the emergence of insomnia\u0026ndash;GS genetic convergence.\u003c/p\u003e \u003cdiv id=\"Sec31\" class=\"Section2\"\u003e \u003ch2\u003eMechanistic insights into insomnia\u0026ndash;GS interactions\u003c/h2\u003e \u003cp\u003eCell-type\u0026ndash;specific TWAS and SMR analyses identified astrocytes as a convergent interface linking insomnia to GS function. Astrocytic \u003cem\u003eMAPT\u003c/em\u003e effects support mechanisms involving altered endfoot polarity, \u003cem\u003eAQP4\u003c/em\u003e mislocalization, or tau-driven disruption of neurovascular coupling(\u003cspan citationid=\"CR52\" class=\"CitationRef\"\u003e52\u003c/span\u003e). \u003cem\u003eHEXIM1\u003c/em\u003e(\u003cspan citationid=\"CR49\" class=\"CitationRef\"\u003e49\u003c/span\u003e) and \u003cem\u003eACBD4\u003c/em\u003e(\u003cspan citationid=\"CR50\" class=\"CitationRef\"\u003e50\u003c/span\u003e) point toward transcriptional and metabolic pathways influencing astrocyte and endothelial contributions to perivascular fluid dynamics. Together, these results outline a model in which insomnia modifies glia-regulated clearance pathways, influencing regional susceptibility to impaired glymphatic transport and potentially facilitating downstream toxic protein accumulation.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec32\" class=\"Section2\"\u003e \u003ch2\u003eLimitations and future directions\u003c/h2\u003e \u003cp\u003eSeveral limitations warrant consideration. First, although ALPS indices provide genetically informative and scalable proxies of GS function, they capture only perivascular diffusivity and do not encompass the full GS pathway\u0026mdash;particularly meningeal lymphatic drainage(\u003cspan citationid=\"CR58\" class=\"CitationRef\"\u003e58\u003c/span\u003e)\u0026mdash;and no gold-standard in vivo metric of GS function currently exists. Second, the insomnia GWAS used in this study is based on self-reported symptoms in UK Biobank, which introduces two layers of unavoidable uncertainty: (i) the primary GWAS does not explicitly exclude individuals with neurodegenerative disorders, raising the possibility that a subset of cases reflects sleep disturbance secondary to early or established neurodegeneration rather than primary insomnia; and (ii) insomnia was defined using questionnaire-based phenotyping rather than clinical diagnosis, which increases heterogeneity and may attenuate trait-specific genetic signals. Although the original GWAS included sensitivity analyses excluding selected chronic and psychiatric illnesses, comprehensive removal of neurodegenerative cases is not feasible using publicly available summary statistics. Third, ST mapping analyses were limited by the lack of ST data from individuals with insomnia. Finally, although we identified convergent genetic pathways linking insomnia and GS function, causal directionality among insomnia, GS dysfunction, and neurodegenerative processes cannot be inferred from cross-sectional GWAS and requires longitudinal and experimental confirmation.\u003c/p\u003e \u003cp\u003eFuture studies combining dynamic sleep recordings, multimodal neuroimaging, CSF biomarkers, and cell-type\u0026ndash;specific perturbations will be essential for elucidating how sleep behavior, glial physiology, and clearance mechanisms interact over time. Such efforts may help identify therapeutic strategies targeting sleep or GS pathways to mitigate neurodegenerative risk.\u003c/p\u003e "},{"header":"Declarations","content":"\u003cp\u003e\u003cstrong\u003eDATA AVAILABILITY\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eInsomnia GWAS: publicly available at the GWAS Catalog (GCST007387), https://www.ebi.ac.uk/gwas/studies/GCST007387.\u003c/p\u003e\n\u003cp\u003eALPS GWAS: available via the GWAS Catalog under publications 39823331 and 39805841. https://www.ebi.ac.uk/gwas/publications/39823331. https://www.ebi.ac.uk/gwas/publications/39805841.\u003c/p\u003e\n\u003cp\u003eGTEx v8 eQTLs (for SMR and TWAS): https://yanglab.westlake.edu.cn/software/smr/#eQTLsummarydata.\u003c/p\u003e\n\u003cp\u003eCell-type\u0026ndash;specific cis-eQTLs: https://zenodo.org/records/7276971.\u003c/p\u003e\n\u003cp\u003eCell type\u0026ndash;specific prediction models for scPrediXcan: https://zenodo.org/records/14346661.\u003c/p\u003e\n\u003cp\u003e1000 Genomes Phase 3 reference (European): available through standard LD reference panels used by MiXeR, LAVA, and SMR\u003c/p\u003e\n\u003cp\u003eMAGMA gene location files: https://ctg.cncr.nl/software/MAGMA/aux_files/NCBI37.3.zip.\u003c/p\u003e\n\u003cp\u003egsMap spatial framework: https://github.com/LieberInstitute/Visium_SPG_AD.\u003c/p\u003e\n\u003cp\u003eAll QC-filtered GWAS files (Insomnia, ALPS), the harmonized datasets used for MiXeR, LAVA, GPA, and PleioFDR analyses, and the processed spatial transcriptomic matrices used for gsMap have been deposited in Figshare and are accessible at:\u003c/p\u003e\n\u003cp\u003ehttps://doi.org/10.6084/m9.figshare.30751190.\u003c/p\u003e\n\u003cp\u003eAll data used in the manuscript are publicly available, and no individual-level confidential data were used.\u003c/p\u003e\n\n\u003cp\u003e\u003cstrong\u003eSOFTWARE USED\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eAll analyses were performed using publicly available software. The key tools include:\u003c/p\u003e\n\u003cp\u003eGPA (v1.2.1), https://dongjunchung.github.io/GPA/\u003c/p\u003e\n\u003cp\u003eMiXeR (v2.2.1), https://github.com/precimed/mixer\u003c/p\u003e\n\u003cp\u003eLAVA (v0.1.0), https://github.com/josefin-werme/LAVA\u003c/p\u003e\n\u003cp\u003eGWAS-PW, https://github.com/joepickrell/gwas-pw\u003c/p\u003e\n\u003cp\u003ePleioFDR (condFDR \u0026amp; conjFDR), https://github.com/precimed/pleiofdr\u003c/p\u003e\n\u003cp\u003eMTAG, https://github.com/JonJala/mtag\u003c/p\u003e\n\u003cp\u003eFine-mapping, colocalization, and integrative functional analyses\u003c/p\u003e\n\u003cp\u003eCOLOC, https://github.com/chr1swallace/coloc\u003c/p\u003e\n\u003cp\u003eecholocatoR (v1.1+), https://rajlabmssm.github.io/echolocatoR/index.html\u003c/p\u003e\n\u003cp\u003eFOCUS, https://github.com/mancusolab/ma-focus/wiki\u003c/p\u003e\n\u003cp\u003eSMR \u0026amp; HEIDI, https://yanglab.westlake.edu.cn/software/smr/#SMR\u0026amp;HEIDIanalysis\u003c/p\u003e\n\u003cp\u003eMetaXcan, https://github.com/hakyimlab/MetaXcan\u003c/p\u003e\n\u003cp\u003escPrediXcan, https://github.com/hakyimlab/scPrediXcan\u003c/p\u003e\n\u003cp\u003eCell-type and spatial-functional annotation analyses\u003c/p\u003e\n\u003cp\u003ePCGA, https://pmglab.top/pcga/#/infer_cell\u003c/p\u003e\n\u003cp\u003eGsMap, https://yanglab.westlake.edu.cn/gsmap/document/software\u003c/p\u003e\n\u003cp\u003eclusterProfiler (v4.10+), http://bioconductor.org/packages/release/bioc/html/clusterProfiler.html\u003c/p\u003e\n\u003cp\u003eaPEAR, https://github.com/kerseviciute/aPEAR\u003c/p\u003e\n\u003cp\u003eGeneral genomic annotation tools\u003c/p\u003e\n\u003cp\u003eFUMA, https://fuma.ctglab.nl/\u003c/p\u003e\n\u003cp\u003eAll additional R and Python packages used in preprocessing, visualization, and statistical analyses are standard and listed within the GitHub repository described in the \u0026ldquo;Code Availability\u0026rdquo; section.\u003c/p\u003e\n\n\u003cp\u003e\u003cstrong\u003eCODE AVAILABILITY\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eAll scripts used for preprocessing, QC, statistical analyses, and figure generation are based on the publicly available software and repositories cited in the \u0026ldquo;Software Availability\u0026rdquo; section.\u003c/p\u003e\n\u003cp\u003eThe custom code used to reproduce the analytical workflow (including data harmonization, MiXeR fitting pipelines, LAVA block-level analyses, multi-method fine-mapping, conditional FDR procedures, TWAS/MetaXcan models, cell-type mapping, and gsMap integration) has been deposited in GitHub: https://github.com/junkman666/Insomnia.git\u003c/p\u003e\u003cp\u003e\u003cstrong\u003eACKNOWLEDGEMENTS\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eWe thank the research teams who generated and openly shared the GWAS, eQTL, cell-type\u0026ndash;specific, and ST datasets that made this study possible. We are grateful to the developers and maintainers of the publicly available tools used in this work, including MiXeR, LAVA, GPA, PleioFDR, GWAS-PW, COLOC, FOCUS, MetaXcan, scPrediXcan, PCGA, GsMap, and associated computational pipelines.\u003c/p\u003e\n\n\u003cp\u003e\u003cstrong\u003eAUTHOR CONTRIBUTIONS\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eJ.S., Y.S. and C.X. designed and supervised the study.\u003c/p\u003e\n\u003cp\u003eJ.S. performed data preprocessing, statistical analyses, and integrative genomic modeling. Y.S. and C.X. contributed to methodological refinement, result interpretation, and quality control.\u003c/p\u003e\n\u003cp\u003eAll authors drafted, revised, and approved the final manuscript.\u003c/p\u003e\n\n\u003cp\u003e\u003cstrong\u003eFUNDING\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThis work was supported by Suzhou Science and Technology Bureau Basic research on medical applications -Research on innovative medical applications Project (SKY2023101).\u003c/p\u003e\n\n\u003cp\u003e\u003cstrong\u003eCOMPETING INTERESTS\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe authors declare no competing interests.\u003c/p\u003e\n\n\u003cp\u003e\u003cstrong\u003eETHICS APPROVAL\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThis study used only publicly available, de-identified summary-level genomic and transcriptomic data. No new human participant recruitment, intervention, or identifiable information was involved. Therefore, ethics approval was not required.\u003c/p\u003e\n\n\u003cp\u003e\u003cstrong\u003eINFORMED CONSENT\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eAll datasets analyzed in this study were obtained from public repositories with existing ethics approval and participant informed consent procedures. No new consent was required for the reuse of anonymized summary data.\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\u003cli\u003e\u003cspan\u003eAdamantidis, A.R., de Lecea, L.: Sleep and the hypothalamus. Science. \u003cb\u003e382\u003c/b\u003e, 405\u0026ndash;412 (2023)\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eZou, Q., Zou, G., Wang, S., Wang, Y., Xu, J., Long, Y., et al.: Cortical hierarchy underlying homeostatic sleep pressure alleviation. Nat. Commun. \u003cb\u003e16\u003c/b\u003e, 10014 (2025)\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eChaput, J.-P., McHill, A.W., Cox, R.C., Broussard, J.L., Dutil, C., da Costa, B.G.G., et al.: The role of insufficient sleep and circadian misalignment in obesity. Nat. Rev. Endocrinol. \u003cb\u003e19\u003c/b\u003e, 82\u0026ndash;97 (2023)\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eZhang, X., Yin, J., Sun, X., Qu, Z., Zhang, J., Zhang, H.: The association between insomnia and cognitive decline: A scoping review. Sleep. Med. \u003cb\u003e124\u003c/b\u003e, 540\u0026ndash;550 (2024)\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003ePerlis, M.L., Posner, D., Riemann, D., Bastien, C.H., Teel, J., Thase, M.: Insomnia. Lancet Lond. Engl. \u003cb\u003e400\u003c/b\u003e, 1047\u0026ndash;1060 (2022)\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBlanco-Garc\u0026iacute;a, F.J., L\u0026oacute;pez-L\u0026oacute;pez, J.A., Ordo\u0026ntilde;ana, J.R., Madrid-Valero, J.J.: Genetic association between sleep quality, insomnia, and psychological distress: a systematic review and meta-analysis. Sleep. Med. Rev. \u003cb\u003e83\u003c/b\u003e, 102149 (2025)\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMadrid-Valero, J.J., Gregory, A.M.: Behaviour genetics and sleep: A narrative review of the last decade of quantitative and molecular genetic research in humans. Sleep. Med. Rev. \u003cb\u003e69\u003c/b\u003e, 101769 (2023)\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSonti, S., Grant, S.F.A.: Leveraging genetic discoveries for sleep to determine causal relationships with common complex traits. Sleep. \u003cb\u003e45\u003c/b\u003e, zsac180 (2022)\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKozhemiako, N., Mita, C., Panagiotaropoulou, G.M., Lewis, K.J.S., Sofer, T., Purcell, S.M.: The Interplay Between Sleep and Mental Health: A Genetic Perspective. Biol. Psychiatry. \u003cb\u003e98\u003c/b\u003e, 863\u0026ndash;874 (2025)\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eChong, P.L.H., Garic, D., Shen, M.D., Lundgaard, I., Schwichtenberg, A.J.: Sleep, cerebrospinal fluid, and the glymphatic system: A systematic review. Sleep. Med. Rev. \u003cb\u003e61\u003c/b\u003e, 101572 (2022)\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eReiter, R.J., Sharma, R., Cucielo, M.S., Tan, D.X., Rosales-Corral, S., Gancitano, G., de Almeida Chuffa, L.G.: Brain washing and neural health: role of age, sleep, and the cerebrospinal fluid melatonin rhythm. Cell. Mol. Life Sci. CMLS. \u003cb\u003e80\u003c/b\u003e, 88 (2023)\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eRasmussen, M.K., Mestre, H., Nedergaard, M.: Fluid transport in the brain. Physiol. Rev. \u003cb\u003e102\u003c/b\u003e, 1025\u0026ndash;1151 (2022)\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eIliff, J.J., Wang, M., Liao, Y., Plogg, B.A., Peng, W., Gundersen, G.A., et al.: A paravascular pathway facilitates CSF flow through the brain parenchyma and the clearance of interstitial solutes, including amyloid β. Sci. Transl Med. \u003cb\u003e4\u003c/b\u003e, 147ra111 (2012)\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHuang, S.-Y., Ge, Y.-J., Ren, P., Wu, B.-S., Gong, W., Du, J., et al.: Genome-wide association study unravels mechanisms of brain glymphatic activity. Nat. Commun. \u003cb\u003e16\u003c/b\u003e, 626 (2025)\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eRan, L., Fang, Y., Cheng, C., He, Y., Shao, Z., Kong, Y., et al.: Genome-wide and phenome-wide studies provided insights into brain glymphatic system function and its clinical associations. Sci. Adv. \u003cb\u003e11\u003c/b\u003e, eadr4606 (2025)\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eDivecha, Y.A., Rampes, S., Tromp, S., Boyanova, S.T., Fleckney, A., Fidanboylu, M., Thomas, S.A.: The microcirculation, the blood-brain barrier, and the neurovascular unit in health and Alzheimer disease: The aberrant pericyte is a central player. Pharmacol. Rev. \u003cb\u003e77\u003c/b\u003e, 100052 (2025)\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLane, J.M., Jones, S.E., Dashti, H.S., Wood, A.R., Aragam, K.G., van Hees, V.T., et al.: Biological and clinical insights from genetics of insomnia symptoms. Nat. Genet. \u003cb\u003e51\u003c/b\u003e, 387\u0026ndash;393 (2019)\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKwon, S.H., Parthiban, S., Tippani, M., Divecha, H.R., Eagles, N.J., Lobana, J.S., et al.: Influence of alzheimer\u0026rsquo;s disease related neuropathology on local microenvironment gene expression in the human inferior temporal cortex. GEN. Biotechnol. \u003cb\u003e2\u003c/b\u003e, 399\u0026ndash;417 (2023)\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eChen, A., Liao, S., Cheng, M., Ma, K., Wu, L., Lai, Y., et al.: Spatiotemporal transcriptomic atlas of mouse organogenesis using DNA nanoball-patterned arrays. Cell. \u003cb\u003e185\u003c/b\u003e, 1777\u0026ndash;1792e21 (2022)\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHolland, D., Frei, O., Desikan, R., Fan, C.-C., Shadrin, A.A., Smeland, O.B., et al.: Beyond SNP heritability: Polygenicity and discoverability of phenotypes estimated with a univariate Gaussian mixture model. PLoS Genet. \u003cb\u003e16\u003c/b\u003e, e1008612 (2020)\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eFrei, O., Holland, D., Smeland, O.B., Shadrin, A.A., Fan, C.C., Maeland, S., et al.: Bivariate causal mixture model quantifies polygenic overlap between complex traits beyond genetic correlation. Nat. Commun. \u003cb\u003e10\u003c/b\u003e, 2417 (2019)\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eFrei, O., Hindley, G., Shadrin, A.A., van der Meer, D., Akdeniz, B.C., Hagen, E., et al.: Improved functional mapping of complex trait heritability with GSA-MiXeR implicates biologically specific gene sets. Nat. Genet. \u003cb\u003e56\u003c/b\u003e, 1310\u0026ndash;1318 (2024)\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eChung, D., Yang, C., Li, C., Gelernter, J., Zhao, H.: GPA: a statistical approach to prioritizing GWAS results by integrating pleiotropy and annotation ((H. Tang, editor)). \u003cem\u003ePLoS Genet\u003c/em\u003e 10: e1004787. (2014)\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eWerme, J., van der Sluis, S., Posthuma, D., de Leeuw, C.A.: An integrated framework for local genetic correlation analysis. Nat. Genet. \u003cb\u003e54\u003c/b\u003e, 274\u0026ndash;282 (2022)\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eZeng, R., Jiang, R., Huang, W., Wang, J., Zhang, L., Ma, Y., et al.: Dissecting shared genetic architecture between obesity and multiple sclerosis. EBioMedicine. \u003cb\u003e93\u003c/b\u003e, 104647 (2023)\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSmeland, O.B., Frei, O., Shadrin, A., O\u0026rsquo;Connell, K., Fan, C.-C., Bahrami, S., et al.: Discovery of shared genomic loci using the conditional false discovery rate approach. Hum. Genet. \u003cb\u003e139\u003c/b\u003e, 85\u0026ndash;94 (2020)\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eAndreassen, O.A., Thompson, W.K., Schork, A.J., Ripke, S., Mattingsdal, M., Kelsoe, J.R., et al.: Improved detection of common variants associated with schizophrenia and bipolar disorder using pleiotropy-informed conditional false discovery rate. PLoS Genet. \u003cb\u003e9\u003c/b\u003e, e1003455 (2013)\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eWallace, C.: A more accurate method for colocalisation analysis allowing for multiple causal variants. PLoS Genet. \u003cb\u003e17\u003c/b\u003e, e1009440 (2021)\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eWang, G., Sarkar, A., Carbonetto, P., Stephens, M.: A simple new approach to variable selection in regression, with application to genetic fine mapping. J. R Stat. Soc. Ser. B Stat. Methodol. \u003cb\u003e82\u003c/b\u003e, 1273\u0026ndash;1300 (2020)\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBenner, C., Spencer, C.C.A., Havulinna, A.S., Salomaa, V., Ripatti, S., Pirinen, M.: FINEMAP: efficient variable selection using summary data from genome-wide association studies. Bioinforma Oxf. Engl. \u003cb\u003e32\u003c/b\u003e, 1493\u0026ndash;1501 (2016)\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSchilder, B.M., Humphrey, J., Raj, T.: echolocatoR: an automated end-to-end statistical and functional genomic fine-mapping pipeline. Bioinforma Oxf. Engl. \u003cb\u003e38\u003c/b\u003e, 536\u0026ndash;539 (2022)\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMancuso, N., Freund, M.K., Johnson, R., Shi, H., Kichaev, G., Gusev, A., Pasaniuc, B.: Probabilistic fine-mapping of transcriptome-wide association studies. Nat. Genet. \u003cb\u003e51\u003c/b\u003e, 675\u0026ndash;682 (2019)\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBerisa, T., Pickrell, J.K.: Approximately independent linkage disequilibrium blocks in human populations. Bioinforma Oxf. Engl. \u003cb\u003e32\u003c/b\u003e, 283\u0026ndash;285 (2016)\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eShi, H., Burch, K.S., Johnson, R., Freund, M.K., Kichaev, G., Mancuso, N., et al.: Localizing Components of Shared Transethnic Genetic Architecture of Complex Traits from GWAS Summary Data. Am. J. Hum. Genet. \u003cb\u003e106\u003c/b\u003e, 805\u0026ndash;817 (2020)\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eWatanabe, K., Taskesen, E., van Bochoven, A., Posthuma, D.: Functional mapping and annotation of genetic associations with FUMA. Nat. Commun. \u003cb\u003e8\u003c/b\u003e, 1826 (2017)\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eWang, K., Li, M., Hakonarson, H.: ANNOVAR: functional annotation of genetic variants from high-throughput sequencing data. Nucleic Acids Res. \u003cb\u003e38\u003c/b\u003e, e164 (2010)\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBarbeira, A.N., Dickinson, S.P., Bonazzola, R., Zheng, J., Wheeler, H.E., Torres, J.M., et al.: Exploring the phenotypic consequences of tissue specific gene expression variation inferred from GWAS summary statistics. Nat. Commun. \u003cb\u003e9\u003c/b\u003e, 1825 (2018)\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBarbeira, A.N., Pividori, M., Zheng, J., Wheeler, H.E., Nicolae, D.L., Im, H.K.: Integrating predicted transcriptome from multiple tissues improves association detection. PLoS Genet. \u003cb\u003e15\u003c/b\u003e, e1007889 (2019)\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eGTEx Consortium: The GTEx Consortium atlas of genetic regulatory effects across human tissues. Science. \u003cb\u003e369\u003c/b\u003e, 1318\u0026ndash;1330 (2020)\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eZhu, Z., Zhang, F., Hu, H., Bakshi, A., Robinson, M.R., Powell, J.E., et al.: Integration of summary data from GWAS and eQTL studies predicts complex trait gene targets. Nat. Genet. \u003cb\u003e48\u003c/b\u003e, 481\u0026ndash;487 (2016)\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBryois, J., Calini, D., Macnair, W., Foo, L., Urich, E., Ortmann, W., et al.: Cell-type-specific cis-eQTLs in eight human brain cell types identify novel risk genes for psychiatric and neurological disorders. Nat. Neurosci. \u003cb\u003e25\u003c/b\u003e, 1104\u0026ndash;1112 (2022)\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003e23andMe Research Team, Social Science Genetic Association Consortium, Turley, P., Walters, R.K., Maghzian, O., Okbay, A., et al.: Multi-trait analysis of genome-wide association summary statistics using MTAG. Nat. Genet. \u003cb\u003e50\u003c/b\u003e, 229\u0026ndash;237 (2018)\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eYang, Y., Chen, J., Zhao, X., Gong, F., Liu, R., Miao, J., et al.: Genetic analysis reveals the shared genetic architecture between breast cancer and atrial fibrillation. Front. Genet. \u003cb\u003e16\u003c/b\u003e, 1450259 (2025)\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSong, L., Chen, W., Hou, J., Guo, M., Yang, J.: Spatially resolved mapping of cells associated with human complex traits. Nature. \u003cb\u003e641\u003c/b\u003e, 932\u0026ndash;941 (2025)\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eStefani, A., Antelmi, E., Arnaldi, D., Arnulf, I., During, E., H\u0026ouml;gl, B., et al.: From mechanisms to future therapy: a synopsis of isolated REM sleep behavior disorder as early synuclein-related disease. Mol. Neurodegener. \u003cb\u003e20\u003c/b\u003e, 19 (2025)\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eGoh, S., Thiyagarajan, L., Dudding-Byth, T., Pinese, M., Kirk, E.P.: A systematic review and pooled analysis of penetrance estimates of copy-number variants associated with neurodevelopment. Genet. Med. Off J. Am. Coll. Med. Genet. \u003cb\u003e27\u003c/b\u003e, 101227 (2025)\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003ePedicone, C., Weitzman, S.A., Renton, A.E., Goate, A.M.: Unraveling the complex role of MAPT-containing H1 and H2 haplotypes in neurodegenerative diseases. Mol. Neurodegener. \u003cb\u003e19\u003c/b\u003e, 43 (2024)\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eShaw-Smith, C., Pittman, A.M., Willatt, L., Martin, H., Rickman, L., Gribble, S., et al.: Microdeletion encompassing MAPT at chromosome 17q21.3 is associated with developmental delay and learning disability. Nat. Genet. \u003cb\u003e38\u003c/b\u003e, 1032\u0026ndash;1037 (2006)\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eFujinaga, K., Huang, F., Peterlin, B.M.: P-TEFb: The master regulator of transcription elongation. Mol. Cell. \u003cb\u003e83\u003c/b\u003e, 393\u0026ndash;403 (2023)\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eCostello, J.L., Koster, J., Silva, B.S.C., Worthy, H.L., Schrader, T.A., Hacker, C., et al.: Differential roles for ACBD4 and ACBD5 in peroxisome-ER interactions and lipid metabolism. J. Biol. Chem. \u003cb\u003e299\u003c/b\u003e, 105013 (2023)\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBeauchamp, M.-C., Jerome-Majewska, L.A.: A protective role for EFTUD2 in the brain. Neuron. \u003cb\u003e112\u003c/b\u003e, 3378\u0026ndash;3380 (2024)\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eIshida, K., Yamada, K., Nishiyama, R., Hashimoto, T., Nishida, I., Abe, Y., et al.: Glymphatic system clears extracellular tau and protects from tau aggregation and neurodegeneration. J. Exp. Med. \u003cb\u003e219\u003c/b\u003e, e20211275 (2022)\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eJansen, P.R., Watanabe, K., Stringer, S., Skene, N., Bryois, J., Hammerschlag, A.R., et al.: Genome-wide analysis of insomnia in 1,331,010 individuals identifies new risk loci and functional pathways. Nat. Genet. \u003cb\u003e51\u003c/b\u003e, 394\u0026ndash;403 (2019)\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eDoherty, A., Smith-Byrne, K., Ferreira, T., Holmes, M.V., Holmes, C., Pulit, S.L., Lindgren, C.M.: GWAS identifies 14 loci for device-measured physical activity and sleep duration. Nat. Commun. \u003cb\u003e9\u003c/b\u003e, 5257 (2018)\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eNagelhus, E.A., Ottersen, O.P.: Physiological roles of aquaporin-4 in brain. Physiol. Rev. \u003cb\u003e93\u003c/b\u003e, 1543\u0026ndash;1562 (2013)\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003ePinho-Correia, L.M., McCullough, S.J.C., Ghanizada, H., Nedergaard, M., Rustenhoven, J., Da Mesquita, S.: CSF transport at the brain-meningeal border: effects on neurological health and disease. Lancet Neurol. \u003cb\u003e24\u003c/b\u003e, 535\u0026ndash;547 (2025)\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLi, X., Liu, G., Yang, L., Li, Z., Zhang, Z., Xu, Z., et al.: Decoding Cortical Glial Cell Development. Neurosci. Bull. \u003cb\u003e37\u003c/b\u003e, 440\u0026ndash;460 (2021)\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKeil, S.A., Jansson, D., Braun, M., Iliff, J.J.: Glymphatic dysfunction in Alzheimer\u0026rsquo;s disease: A critical appraisal. Science. \u003cb\u003e389\u003c/b\u003e, eadv8269 (2025)\u003c/span\u003e\u003c/li\u003e\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":false,"highlight":"","institution":"","isAcceptedByJournal":false,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"nature-portfolio","isNatureJournal":true,"hasQc":false,"allowDirectSubmit":false,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"","title":"Nature Portfolio","twitterHandle":"","acdcEnabled":false,"dfaEnabled":false,"editorialSystem":"ejp","reportingPortfolio":"","inReviewEnabled":true,"inReviewRevisionsEnabled":false},"keywords":"Insomnia, Glymphatic system, DTI-ALPS, Pleiotropy, 17q21.31 locus, Integrative genomics","lastPublishedDoi":"10.21203/rs.3.rs-8585990/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-8585990/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003eSleep plays a critical role in brain waste clearance, yet whether insomnia shares a genetic basis with the glymphatic system (GS)\u0026mdash;a glia-dependent perivascular pathway involved in metabolite removal\u0026mdash;remains unclear. Here, we integrated large-scale genome-wide association studies (GWAS) of insomnia with diffusion tensor imaging along the perivascular space (DTI\u0026ndash;ALPS), an imaging-derived proxy of GS function, across two independent cohorts. Polygenic enrichment analyses revealed localized genetic sharing between insomnia and multiple ALPS phenotypes despite minimal genome-wide genetic correlation. Conjunctional false discovery rate and Bayesian colocalization analyses identified shared causal signals at 17q21.31, a pleiotropic locus encompassing the MAPT inversion region. Transcriptome-wide association, gene-level fine-mapping, and summary-data Mendelian randomization converged on \u003cem\u003eHEXIM1\u003c/em\u003e, \u003cem\u003eACBD4\u003c/em\u003e, \u003cem\u003eEFTUD2\u003c/em\u003e, and \u003cem\u003eMAPT\u003c/em\u003e as shared genes influencing both insomnia and GS function. Functional characterization showed that these genes were enriched across multiple brain regions and cell types, including neurons, astrocytes, microglia, oligodendroglia, and vascular-associated cells. Notably, gene-level effects exhibited regional and phenotype-specific heterogeneity. Together, our findings demonstrate that insomnia and glymphatic function converge through a context-dependent genetic architecture centered on 17q21.31, implicating neuroglial pathways relevant to protein clearance and Alzheimer\u0026rsquo;s disease vulnerability.\u003c/p\u003e","manuscriptTitle":"A multilayer genomic framework linking insomnia to glymphatic system function through pleiotropic mechanisms at 17q21.31","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2026-02-25 16:51:17","doi":"10.21203/rs.3.rs-8585990/v1","editorialEvents":[],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"communications-biology","isNatureJournal":true,"hasQc":false,"allowDirectSubmit":false,"externalIdentity":"commsbio","sideBox":"Learn more about [Communications Biology](http://www.nature.com/commsbio/)","snPcode":"","submissionUrl":"","title":"Communications Biology","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"ejp","reportingPortfolio":"Communications Series","inReviewEnabled":true,"inReviewRevisionsEnabled":false}}],"origin":"","ownerIdentity":"d96a8c0f-96d6-45d9-a3b6-5685643c1979","owner":[],"postedDate":"February 25th, 2026","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"under-review","subjectAreas":[{"id":63255618,"name":"Biological sciences/Genetics/Genetic association study/Genome-wide association studies"},{"id":63255619,"name":"Health sciences/Neurology/Neurological disorders/Sleep disorders"}],"tags":[],"updatedAt":"2026-04-14T03:50:37+00:00","versionOfRecord":[],"versionCreatedAt":"2026-02-25 16:51:17","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-8585990","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-8585990","identity":"rs-8585990","version":["v1"]},"buildId":"XKTyCvWXoU3ODBz1xrDgd","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}