Binhe Yu

A growing body of evidence supports the comorbidity between depression (DEP) and obesity, yet the genetic mechanisms underlying this association remain unclear. Our study explored the shared genetic architecture and causal associations of DEP with BMI. We investigated the multigene overlap and genetic correlation between DEP (N > 1.3 million) and BMI (N = 806,834) based on genome-wide association studies (GWAS) and using the bivariate causal mixture model and linkage disequilibrium score regression (LDSC). The causal association was explored by bi-directional Mendelian randomization (MR). Common risk loci were identified through cross-trait meta-analyses. Stratified LDSC and multi-marker gene annotation analyses were applied to investigate single-nucleotide polymorphisms enrichment across tissue types, cell types, and functional categories. Finally, we explored shared functional genes by Summary Data-Based Mendelian Randomization (SMR) and further detected differential expression genes (DEG) in brain tissues of individuals with depression and obesity. We found a positive genetic correlation between DEP and BMI (rg = 0.19, P = 4.07 × 10−26), which was more evident in local genomic regions. Cross-trait meta-analyses identified 16 shared genetic loci, 5 of which were newly identified, and they had influence on both diseases in the same direction. MR analysis showed a bidirectional causal association between DEP and BMI, with comparable effect sizes estimated in both directions. Combined with gene expression information, we found that genetic correlations between DEP and BMI were enriched in 6 brain regions, predominantly in the nucleus accumbens and anterior cingulate cortex. Moreover, 6 specific cell types and 23 functional genes were found to have an impact on both DEP and BMI across the brain regions. Of which, NEGR1 was identified as the most significant functional gene and associated with DEP and BMI at the genome-wide significance level (P < 5 × 10−8). Compared with healthy controls, the expression levels of NEGR1 gene were significant lower in brain tissues of individuals with depression and obesity. Our study reveals shared genetic basis underpinnings between DEP and BMI, including genetic correlations and common genes. These insights offer novel opportunities and avenues for future research into their comorbidities.

Abstract Background Despite a strong link between obstructive sleep apnea (OSA) and sleep traits, the shared genetic architecture remains unclear. This study aims to explore the shared genetic basis and bidirectional causal between OSA and sleep traits. Methods Using large‐scale genome‐wide association studies summary statistics for OSA and sleep traits, we employed linkage disequilibrium score regression (LDSR) and MiXeR to examine genetic correlations and quantify polygenic overlaps. The causal association was explored by bidirectional Mendelian randomization. In addition, we identified shared genomic loci through conditional and conjunctional false discovery rate (cond/conjFDR) analysis, followed by annotation to identify shared genes. Finally, we performed enrichment, developmental trajectory, and phenome‐wide association study analysis of the shared genes to explore underlying mechanisms. Results We found that both LDSR and MiXeR results revealed substantial genetic correlations and polygenic overlaps between OSA and most of sleep traits. MR analysis supported bidirectional causality between OSA and sleep traits such as insomnia and snoring. Subsequent conjFDR analysis pinpointed 168 distinct shared loci, which encompassed 695 unique genes, and these genes are predominantly enriched in the neurodevelopmental and metabolic process pathways. Notably, the expression of 38 shared genes exhibits a significant correlation with both OSA and sleep traits. These shared genes exhibit specific developmental trajectories and demonstrate significant pleiotropic associations with phenotypes such as metabolism, immunity, and brain structure. Conclusion This study uncovers the broad pleiotropy of the genetic architecture shared between OSA and sleep traits, highlighting neurodevelopmental and metabolic pathways as the key shared biological underpinnings.

Backgrounds: Observational research has shown significant associations between inflammatory factors and sleep. Experimental studies suggested acute increase in the levels of inflammatory markers following sleep deprivation and sleep restriction. However, the causal association between inflammatory factors and sleep remains unclear in chronic and natural settings. Objectives: This study aimed to investigate the causal association of inflammatory factors with chronotype, daytime napping, daytime sleepiness, insomnia symptoms, and sleep duration. Methods: Two-sample bidirectional Mendelian randomization (MR) analysis was employed to investigate the causal associations between 91 inflammatory factors and 7 sleep-related traits. Summary-level data of inflammatory factors were derived from the EBI GWAS Catalog (n = 14,824); sleep-related traits were obtained from UK Biobank. We calculated effect estimates using the inverse-variance weighted (IVW), weighted median, and MR-Egger methods. Heterogeneity and pleiotropy were detected and measured by the MR pleiotropy residual sum and outlier, Cochran's Q statistics, and MR-Egger regression. Results: Significant bidirectional causal associations were observed. The most crucial findings included the causal effects of CD40 (OR = 1.02, 95 % CI: 1.01-1.03), ST1A1 (OR = 0.97, 95 % CI: 0.96-0.99), uPA (OR = 1.03, 95 % CI: 1.01-1.04) on chronotype, and FGF-21 (OR = 1.02, 95 % CI: 1.01-1.03), hGDNF (OR = 1.01, 95 % CI: 1.00-1.02), TNFB (OR = 0.99, 95 % CI: 0.98-1.00), TNFSF14 (OR = 1.01, 95 % CI: 1.00-1.02) on napping. Overall, 30 inflammatory factors were found to causally affect sleep traits, and 20 reciprocal effects were observed. Conclusion: Our study suggested a bidirectional causal association between inflammatory factors and sleep-related traits, such as the roles of CD40, ST1A1, and uPA in regulating chronotype, and FGF-21, hGDNF, and TNFB in influencing daytime napping.

To tackle the challenges of uneven haze distribution, blurring, and color distortion in complex scenes, this paper proposes an unsupervised image dehazing algorithm based on dual discriminators and bidirectional contrastive reconstruction. The framework integrates an end-to-end dehazing network with a parameter estimation module, featuring sub-networks for atmospheric light and transmission map estimation. Leveraging parallel attention, multi-scale learning, and a variational autoencoder, the model effectively captures key features and adaptively estimates haze parameters, enhancing clarity and realism. Experiments on benchmark datasets demonstrate that the proposed method outperforms existing unsupervised approaches in both visual quality and quantitative metrics, while also showcasing strong robustness and practical applicability.

Multi-subject customization aims to synthesize multiple user-specified subjects into a coherent image. To address issues such as subjects missing or conflicts, recent works incorporate layout guidance to provide explicit spatial constraints. However, existing methods still struggle to balance three critical objectives: text alignment, subject identity preservation, and layout control, while the reliance on additional training further limits their scalability and efficiency. In this paper, we present AnyMS, a novel training-free framework for layout-guided multi-subject customization. AnyMS leverages three input conditions: text prompt, subject images, and layout constraints, and introduces a bottom-up dual-level attention decoupling mechanism to harmonize their integration during generation. Specifically, global decoupling separates cross-attention between textual and visual conditions to ensure text alignment. Local decoupling confines each subject's attention to its designated area, which prevents subject conflicts and thus guarantees identity preservation and layout control. Moreover, AnyMS employs pre-trained image adapters to extract subject-specific features aligned with the diffusion model, removing the need for subject learning or adapter tuning. Extensive experiments demonstrate that AnyMS achieves state-of-the-art performance, supporting complex compositions and scaling to a larger number of subjects.