Search Results (396)

Objectives: Individuals with type 2 diabetes (T2D) remain at high cardiovascular disease (CVD) risk. The pan-immune-inflammation value (PIV) integrates circulating neutrophils, monocytes, platelets, and lymphocytes but does not capture albumin-related nutritional and inflammatory reserve. We investigated associations between the PIV-to-albumin ratio (PIVA) and incident CVD, cardiovascular mortality, and disease progression in individuals with T2D. Methods: This prospective study included 15,355 UK Biobank participants with T2D and no baseline CVD. PIVA was calculated from peripheral blood cell counts and serum albumin and was natural log-transformed. Fine–Gray competing-risk and cause-specific Cox models were used to evaluate incident CVD and cardiovascular mortality. Multi-state models characterized transitions from T2D to CVD and death. We also examined nonlinear associations, renal biomarker mediation, joint associations with the CVD polygenic risk score and triglyceride–glucose index, sensitivity analyses, and external validation of cardiovascular mortality in the National Health and Nutrition Examination Survey. Results: During median follow-ups of 12.94 years for incident CVD and 14.49 years for cardiovascular mortality, 4829 incident CVD events and 514 cardiovascular deaths occurred. Each 1-unit increase in lnPIVA was associated with higher risks of incident CVD (subdistribution hazard ratio [sHR], 1.140; 95% confidence interval [CI], 1.088–1.194) and cardiovascular mortality (sHR, 1.449; 95% CI, 1.247–1.684). Associations were nonlinear. Higher lnPIVA was also associated with transitions from T2D to CVD, from T2D directly to cardiovascular death, and from incident CVD to cardiovascular death. Cystatin C explained a larger proportion of these associations than creatinine. Elevated lnPIVA identified excess cardiovascular risk across strata of genetic susceptibility and insulin resistance, and the findings were generally supported by sensitivity and external validation analyses. Conclusions: lnPIVA was associated with incident CVD, cardiovascular mortality, and adverse cardiovascular transitions in individuals with T2D. As an immune-inflammatory and albumin-based index, PIVA may help identify high-risk individuals beyond conventional cardiometabolic and genetic risk profiles. Full article

Background: Research in behavioral genetics has demonstrated that genetic factors significantly contribute to individual differences in behavior, while environmental exposures shape gene expression through epigenetic mechanisms. This interaction is also relevant to the development of antisocial behavior and psychopathic traits. One of the first evidence of gene–environment interaction was the association between low-activity variants of the MAOA gene, childhood maltreatment, and increased risk of antisocial behavior [¹]. Similarly, our research in incarcerated populations showed that adverse paternal parenting is associated with higher levels of psychopathy and the HTR1B rs13212041 TT genotype appears to modulate the individual susceptibility to negative experiences [²]. Single genetic variants, however, exert only modest effects and current evidence supports a polygenic model in which multiple genetic factors interact with environmental adversity to influence neurodevelopment and behavioral outcomes. Using a genome-wide/endophenotype informed analysis, for example, we identified novel gene–environment interactions as risk factors for psychopathy, involving three independent genetic loci in interaction with paternal maltreatment, which were previously associated with disruptive behavior, temperament, and neuroticism [³]. More recently, we also evaluated whether machine-learning models, integrating behavioral, environmental, and genetic variables, could be helpful to predict psychopathic traits. Methods: We compared logistic regression, random forest, support vector machine, XGBoost, and multilayer perceptron. Results: Support vector machine showed the highest accuracy for predicting Psychopathy Check List-Revised (PCL-R) Factor 2 (antisocial lifestyle). Feature-importance analyses identified impulsivity (BIS-11), empathy (IRI), childhood maltreatment (MOPS), and 12 SNPs as the most informative predictors. Notably, removing genetic variables or MOPS scores substantially reduced the model accuracy, indicating that both genetic and environmental information meaningfully contributed to prediction of antisocial behavior. Conclusions: These findings confirm that genetic influences are neither deterministic nor sufficient to explain criminal behavior but may contribute to interindividual differences in vulnerability, particularly through their interaction with environmental and psychosocial factors. In forensic psychiatry, the integration of genetic and environmental information into behavioral assessment may provide additional objective correlates that complement, rather than replace, traditional clinical and psychosocial evaluations. Such an integrated approach could potentially contribute to a more comprehensive understanding of individual vulnerability and behavioral trajectories. However, the use of genetic information in assessments of criminal responsibility should be approached with caution and proven expertise, given the complex, multifactorial nature of antisocial and criminal behavior. Full article

Growth and body conformation traits are key determinants of meat production efficiency and economic performance in sheep. However, the genetic architecture underlying these complex traits remains incompletely understood, particularly across multi-breed populations. In this study, we performed a genome-wide association study (GWAS) for seven growth and developmental traits in a combined population of 401 sheep, including Qira Black, Kyrgyz, Dorset × Hu crossbred, and Suffolk × Karakul crossbred sheep. After genotype harmonization and quality control, 47,674 autosomal SNPs were retained for analysis. Population structure was assessed using principal component analysis, and association testing was conducted using a mixed linear model incorporating breed, principal components, and a kinship matrix. A total of 44 independent loci were detected at a nominal significance threshold, encompassing 112 candidate genes. The strongest association was identified for cannon bone circumference near RPS6KA5 (Chr7; p = 1.40 × 10⁻⁷). Several biologically relevant genes involved in osteogenesis, cartilage development, and metabolic regulation were detected, including STEAP3, SLC26A2, PPARGC1B, COL11A1, CALN1, and CITED2. Two genomic regions exhibited pleiotropic effects, which were identified as being associated with multiple traits, suggesting shared genetic regulation of correlated skeletal characteristics. These findings are consistent with a polygenic architecture underlying growth trait in sheep and highlight candidate genomic regions potentially involved in skeletal development and body conformation. Although further validation is required, the identified loci provide preliminary evidence for regions that may influence growth-related phenotypes and offer a reference for future molecular breeding efforts in indigenous and crossbred sheep populations. Full article

Obesity is a complex, multifactorial chronic disease characterized by excessive and/or aberrant adiposity that requires interventions beyond conventional calorie-restriction models. Current clinical guidelines converge on behavioral modification, encompassing nutritional therapy, physical activity, stress reduction, and sleep improvement, as the cornerstone of obesity management, with psychological therapy, pharmacotherapy, and bariatric procedures as adjunctive or escalating options. While traditional “one-size-fits-all” dietary approaches frequently yield suboptimal long-term outcomes, precision nutrition has emerged as a promising strategy for individualized obesity care. This review critically appraises the evidence underlying each pillar of precision nutrition: genetic profiling shows clear utility in monogenic obesity but variable benefit when diet is matched to common polygenic variants; microbiome-targeted studies reveal that the Bacillota to Bacteroidota ratio shows inconsistent associations with obesity across studies, with genus-level and functional alterations offering more reproducible targets; and metabolomic profiling of postprandial glycemic responses has progressed from foundational predictive algorithms to large-scale clinical validation. It also highlights chrono-nutrition, the alignment of food timing with individual chronotype, as an emerging pillar of personalization. Beyond dietary composition, the review addresses how personalized nutrition can mitigate the adverse effects of incretin-based pharmacotherapy and discusses how artificial intelligence can integrate multi-omics and behavioral data into real-time dietary guidance. Future clinical implementation will require overcoming challenges related to data integration, predictive accuracy, and accessibility, paving the way for more effective, evidence-based obesity management. Full article

Background/Objectives: Post-Traumatic Stress Disorder (PTSD) is a complex psychiatric condition with a strong polygenic and stress-related biological basis. Although Genome-Wide Association Studies (GWAS) have acknowledged abundant risk variants, translating these findings into biologically meaningful candidates remains challenging. This study introduces an integrative computational approach from raw file preparation by python-coded application into downstream in-depth silico analyses designed to systematically refine GWAS signals for PTSD using fine-mapping, linkage disequilibrium (LD), and haplotype analyses. Methods: GWAS source file for PTSD was obtained from the GWAS Catalog (EFO_0001358) and analyzed using a custom Python pipeline integrating data harmonization, genome-wide visualization, LD estimation via 1000 Genomes reference panels, approximate Bayesian fine-mapping, and Haploview-inspired haplotype inference. SNPs were filtered based on statistical significance, LD structure, and posterior inclusion probability. Downstream systems’ biology analyses included protein–protein interaction modeling and pharmacogenomics (PGx) annotations. Results: From the primary GWAS dataset, 100 top-ranked SNPs were selected, leading to the identification of 58 significant loci. LD and haplotype analyses refined these signals to 93 candidate SNPs (66 genes). Following the exclusion of non-protein-coding genes, 45 genes remained, and network-based prioritization generated a final list of 20 biologically connected and pharmacogenetically relevant genes associated with PTSD. Conclusions: This integrative approach provided a robust and reproducible framework for GWAS fine-mapping and variant prioritization, effectively reducing large-scale GWAS outputs to biologically interpretable PTSD risk loci and genes. Full article

Average daily gain (ADG) is a core quantitative trait determining the economic benefits of Ashidan yak, a polled new breed adapted to cold barn feeding on the Qinghai–Tibet Plateau. Unraveling its complex genetic architecture is crucial for early molecular breeding selection. In this study, high-depth whole-genome resequencing (WGS) data from 474 Ashidan yaks were used to conduct combined evaluation of genome-wide association study (GWAS) and genomic selection (GS). During GWAS analysis, sex and measurement batch were included as fixed effects, while birth weight and principal components (PC1–PC3) were incorporated as covariates. Multi-model association analysis using GLM, MLM, and FarmCPU was performed on 3.36 million LD-pruned SNPs. The genomic inflation factors (λ ≈ 1.0) for MLM and FarmCPU confirmed effective elimination of population stratification. A total of 11 genome-wide significant SNP loci and 7 key candidate genes including PDE10A, RAD51B, BCAS3 and KCNH8 were identified via the FarmCPU model. Functional enrichment analysis indicated that these gene clusters are significantly involved in cAMP signaling pathway, regulation of ion channel activity, as well as extracellular matrix remodeling of blood vessels and skeletal muscle cells. For genomic selection, a single-trait GBLUP model was constructed using 22.87 million high-density raw SNPs to fully capture polygenic minor effects. Moderately high narrow-sense genomic heritability of ADG was estimated at h² = 0.3233 (p < 0.05). The average independent prediction accuracy across the 10-fold cross-validation reached an average of R = 0.14 ± 0.07. To evaluate marker prioritized genomic evaluation without data leakage, a strict 10-fold cross-validation scheme was implemented, where the top 1% high-priority variant set (~228,000 SNPs) was screened independently within each training fold. The resulting unbiased prediction accuracy reached R = 0.1328 ± 0.1566 (with an average RMSE of 0.3793 ± 0.0066 and a regression slope of 0.4183 ± 0.5055). Comparing this with the unselected whole-genome baseline (R = 0.14 ± 0.07) indicates that naive marker selection based solely on GBLUP effect size in small reference cohorts is influenced by sampling variance, highlighting the need to integrate multi-omics functional annotations for future custom breeding array development. This study provides quantitative insights into the polygenic architecture of ADG in yaks, offering baseline data for genomic selection and custom array development for indigenous livestock on the Qinghai–Tibet Plateau. Full article

Background/Objectives: The physiological adaptations required to sustain elite football performance are influenced by both genetic background and dynamic biochemical responses, although their interaction across a full competitive season remains insufficiently characterized. This study aimed to examine the association between polygenic profiles and longitudinal biochemical adaptations in professional football players. Methods: Forty male professional football players competing in the Spanish league were monitored across two consecutive seasons. Blood samples were collected at six time points representing different phases of the competitive cycle. Biomarkers related to muscle metabolism, iron status, and hepatic function were analyzed. Polygenic profiles were calculated using Total Genotype Scores (TGS) for muscle performance, hepatic resilience, and metabolic efficiency. Associations were initially explored using Pearson correlations and subsequently evaluated using linear mixed-effects models accounting for repeated measurements within subjects. Results: Exploratory correlation analyses identified several associations between polygenic profiles and biochemical markers. Muscle performance TGS was inversely associated with serum iron (r = −0.36, p = 0.017) and positively associated with CK (r = 0.32, p = 0.041), Hb (r = 0.29, p = 0.046), and Hct (r = 0.33, p = 0.024). Hepatic resilience TGS showed inverse associations with ALT (r = −0.39, p = 0.012), urea (r = −0.51, p = 0.011), and BUN (r = −0.51, p = 0.011). Metabolic efficiency TGS was negatively associated with AST (r = −0.43, p = 0.044), ALT (r = −0.33, p = 0.025), and GGT across multiple time points (p = 0.001–0.013). However, although several nominal associations emerged in linear mixed-effects models accounting for repeated measurements, none remained statistically significant after false discovery rate correction. These findings should therefore be interpreted as exploratory and hypothesis-generating. Conclusions: Polygenic profiles may be associated with inter-individual variability in biochemical adaptations throughout a competitive season. These findings suggest the integration of genomic and biochemical data in precision athlete monitoring, while highlighting causal relationships and predictive applications require further investigation. Full article

Dementia with Lewy Bodies (DLB) is the most common neurodegenerative dementia after Alzheimer’s disease (AD), but it remains challenging to diagnose due to overlapping symptoms and mixed pathologies. This pilot study tested whether DLB has different metabolic blood proteomic profiles compared to AD and controls. Serum was analysed from people with DLB (n = 20), a group with AD (either with Alzheimer’s disease dementia or Mild Cognitive Impairment with a positive amyloid positron emission tomography scan (MCI+/AD) (n = 15), and similarly aged controls (n = 15) using the Olink Metabolism panel encompassing 92 proteins. Six proteins (PILRB, LRIG1, NECTIN2, TINAGL1, SSC4D, and FKBP4) were significantly different in DLB compared with the controls, and one protein (SERPINB8) was differentially expressed when compared with MCI+/AD. Receiver operating characteristic curves for biologically relevant proteins that have previously established roles in neurodegenerative and cognitive properties showed that RNASE3 levels could differentiate DLB from MCI+/AD (area under the curve (AUC) = 0.717, sensitivity = 0.850, and specificity = 0.600). A multimarker model incorporating RNASE3 with phosphorylated tau 217 (pTau217) and polygenic risk scores for LBD achieved improved accuracy in discriminating DLB from MCI+/AD (AUC = 0.963, sensitivity = 1.000, and specificity = 0.900). Pathway analyses revealed dysregulation in cortisol signalling, inflammation resolution, and ErbB4-mediated neuroplasticity, which point towards peripheral protein alterations related to the adaptation to stress, immune regulation, and synaptic integrity in DLB. The present exploratory study highlights potential pathophysiological mechanisms implicated in DLB, suggesting that a multimodal biomarker panel may perform better compared to single proteins. Considering the small sample sizes, the findings will need to be replicated in larger cohorts. Full article

The rapid expansion of next-generation sequencing technologies has generated unprecedented volumes of genomic data; however, translating these data into reliable and clinically actionable insights remains a major challenge in precision medicine. Artificial intelligence (AI) has emerged as a key enabling technology across the genomic medicine pipeline, supporting variant detection, variant interpretation, polygenic risk prediction, disease subtyping, biomarker discovery and treatment–response modelling. This review provides a clinically oriented, pipeline-based synthesis of contemporary AI applications in genomic medicine. Major computational paradigms, including machine learning, deep learning, ensemble methods, multimodal AI, explainable AI frameworks and emerging foundation models, are discussed in the context of their contribution to genomic analysis and clinical decision support. Particular emphasis is placed on the factors that determine model robustness and clinical utility, including dataset composition, class imbalance, label noise, calibration, ancestry representation, distributional shift and external validation. Evidence from rare genetic disorders, cardiovascular genetics and precision oncology is examined to illustrate both successful translational applications and persistent barriers to implementation. The review further analyses common sources of failure in real-world genomic AI systems, including overfitting, limited transportability across populations and sequencing environments, inadequate interpretability, and insufficient prospective validation. Ethical and regulatory challenges are discussed in relation to clinical accountability, genomic privacy, algorithmic bias and equitable implementation. Ultimately, the successful clinical translation of genomic AI will depend not only on methodological innovation, but also on rigorous validation, transparent reporting, continuous calibration, robust governance and sustained expert oversight. Full article

Background: The directional relationship between epigenetic age acceleration (EAA) and musculoskeletal disease remains unresolved. This study integrated bidirectional Mendelian randomization (MR) with multi-layer genomic evidence to evaluate directionality, shared genetic architecture, and robustness to instrument definition. Methods: Four EAA clocks (IEAA, PhenoAA, HannumAA, and GrimAA) and ten musculoskeletal phenotypes were analyzed in a 10 × 4 bidirectional two-sample MR design. EAA instruments underwent GRCh37 functional annotation, genome-wide-significant external-association screening for the index variants and European linkage-disequilibrium proxies, pair-specific Steiger filtering, and conservative Set A/B/C sensitivity analyses. The juvenile-arthritis reverse models underwent instrument-flow reconstruction, strength assessment, liability-scale directionality testing, and minimum-detectable-effect analysis. Additional analyses comprised LD score regression (LDSC), PLACO+ cross-trait locus mapping, Bayesian colocalization, multivariable MR (MVMR) with exact-SNP matched univariable comparators, and integrated evidence synthesis. Results: Forward MR yielded two nominal HannumAA associations. The inverse HannumAA–spondyloarthritis estimate remained directionally consistent across the original, Steiger-filtered, and conservative external-association-filtered sets, whereas the HannumAA–pain-in-thoracic-spine estimate lost nominal significance in the conservative set; no forward result survived correction across 40 tests. GrimAA forward estimates were sensitive to use of the fallback instrument threshold. Reverse MR identified ten nominal associations. For juvenile arthritis, three harmonized instruments had F statistics of 51.25–102.35; liability-scale Steiger comparisons supported the tested direction under all 16 outcome-by-prevalence combinations, although the 788-case discovery GWAS and possible winner’s curse remained important limitations. LDSC identified FDR-significant positive genetic correlations of GrimAA with hip osteoarthritis (r_g = 0.267, p = 8.49 × 10⁻⁵, q = 0.0019) and knee osteoarthritis (r_g = 0.269, p = 9.52 × 10⁻⁵, q = 0.0019). PLACO+ identified 738 genome-wide-significant cross-trait variants and 65 independent loci; six of 37 evaluable loci showed strong colocalization. Of 96 MVMR models, 43 had primary-exposure conditional F ≥ 10, and 32 also had candidate-trait conditional F ≥ 10. After exact-SNP matching, the 43 primary-strength models were operationally classified as 35 partially attenuated and eight independent-signal models, with no fully attenuated model; no adjusted association survived multiplicity correction. Conclusions: The results support a prioritized genomic map with substantial instrument- and model-specific uncertainty. Disease-to-clock signals were richer than clock-to-disease signals, GrimAA shared polygenic architecture with osteoarthritis, and selected loci showed strong shared-variant evidence, while the MR and MVMR findings remained unsuitable for definitive causal or mediation claims. Full article

Atrial fibrillation (AF) is a prevalent cardiac arrhythmia associated with significant morbidity and mortality. Structural remodeling of the left atrium, particularly myocardial fibrosis, plays a key role in AF pathogenesis. Matrix metalloproteinases (MMPs) are critical regulators of extracellular matrix remodeling and may contribute to atrial fibrosis through genetic variation. This case–control study included 179 patients with AF and 56 controls. Eight polymorphisms across five MMP genes (MMP1, MMP2, MMP3, MMP9, and MMP12) were analyzed using PCR-based methods. Associations between single-nucleotide polymorphisms (SNPs), AF susceptibility, recurrence, haplotypes, and gene–gene interactions were assessed. The study population was ethnically homogeneous (Polish), minimizing population stratification bias. No significant differences in allele frequencies were observed between AF and control groups in univariate analysis. However, multivariable logistic regression revealed significant associations for MMP1 rs1799750 and MMP2 rs243864 under recessive inheritance models. Haplotype analysis demonstrated a significant global association with AF (p = 0.027), with specific haplotypes showing markedly increased risk. Multifactor dimensionality reduction identified significant gene–gene interactions, particularly involving SNPs in MMP1, MMP2, MMP3, and MMP12. These findings support a polygenic model of AF susceptibility involving extracellular matrix remodeling pathways and highlight the importance of multi-locus genetic analyses. Full article

Background/Objectives: Substance use behaviors share a complex, overlapping polygenic architecture, yet translating genome-wide association study (GWAS) findings into actionable biological mechanisms remains challenging. This study aimed to characterize the genetic architecture of five substance use traits (alcohol consumption, alcohol dependence, nicotine use, illicit drug use, and behavioral disinhibition) and identify shared and distinct gene expression signatures within the neural circuits governing addiction. Methods: We reanalyzed 7188 individuals from the Minnesota Center for Twin and Family Research (MCTFR) cohort utilizing longitudinal composite phenotypes spanning five substance-use domains and general behavioral disinhibition. Post-QC, 6874 individuals were retained for downstream analysis. Following genomic imputation and linear mixed model GWAS (GEMMA), we utilized the SNipar framework to partition polygenic risk scores (PRS) into direct and indirect genetic effects, investigating intergenerational shifts in genetic penetrance and effects of assortative mating. Finally, we integrated our summary statistics with brain tissue reference panels to perform a transcriptome-wide association study (TWAS) modeling genetically regulated gene expression within neural circuits relevant to addiction. Results: Partitioning of polygenic risk revealed that while surface-level parental DNA correlations were modest (r = 0.08), underlying latent genetic correlations approached unity (R_δ ≈ 0.99), indicating that addiction risk clustering in families is driven by intense assortive mating and concentrated biological inheritance. Multi-phenotype TWAS identified several significant gene–phenotype associations—notably ADAM32 and SLC9A3, which demonstrated pleiotropic effects across multiple substance use categories. Crucially, these significant TWAS signals were enriched in striatal structures (caudate, putamen, substantia nigra) and frontal cortical regions. Conclusions: Our findings support a model of shared genetic liability across diverse substance use behaviors, mediated by specific gene expression patterns in the mesolimbic dopamine system and frontal cortex. By integrating multi-phenotype GWAS and TWAS, this study highlights pleiotropic candidate genes and provides critical insights into the tissue-specific neurobiological pathways underlying addiction vulnerability. Full article

Metabolic dysfunction-associated steatotic liver disease (MASLD) has traditionally been conceptualized as a predominantly metabolic disorder driven by obesity and insulin resistance. However, recent advances in human genetics have revealed a more complex picture that encompasses germline susceptibility variants, protective loss-of-function alleles, polygenic risk models, and somatic clonal evolution. Since the discovery of PNPLA3 (patatin-like phospholipase domain-containing 3) I148M, multiple loci—including TM6SF2, MBOAT7, GCKR, HSD17B13, MTARC1, GPAM, and CIDEB—have substantially expanded the mechanistic understanding of disease heterogeneity and hepatocellular vulnerability. Recent studies integrating partitioned polygenic risk scores and unsupervised phenotypic clustering suggest that MASLD may be organized into at least two predominant subtypes: a liver-specific subtype characterized by intrinsic hepatocellular susceptibility, and a cardiometabolic subtype associated with systemic metabolic dysfunction and increased cardiovascular risk. Analyses of cirrhotic liver tissue have, in turn, demonstrated somatic clonal expansion of hepatocytes harboring adaptive metabolic mutations, adding an evolutionary dimension to advanced disease. On this basis, we propose an integrated LS/CM/C framework encompassing liver-specific (LS), cardiometabolic (CM), and clonal (C) components. This model offers a conceptual structure that links germline genetics, metabolic heterogeneity, somatic adaptation, and emerging pharmacogenomic strategies. The recent development of genotype-directed therapies targeting PNPLA3 and HSD17B13, together with the approval of resmetirom and semaglutide, further supports the transition toward biologically stratified hepatology. Although prospective validation remains necessary, the convergence of genetics, clonal biology, and targeted therapeutics suggests that MASLD is moving toward an era of precision medicine. Full article

Staphylococcus pseudintermedius is a common opportunistic pathogen in companion animals and a leading cause of skin and soft tissue infections (SSTIs). Despite its clinical relevance, the genomic determinants underlying pathogenicity and the transition from commensal carriage to invasive infection remain poorly understood. This study aimed to identify the genomic determinants of SSTI pathogenic potential in S. pseudintermedius and to determine whether pathogenicity is driven by single, major-effect virulence genes or by polygenic genome-wide genetic architecture. Using accessory gene-based and unitig-based genome-wide association studies (GWASs), employing a linear mixed model, across a genetically diverse collection of S. pseudintermedius isolates spanning multiple phylogenetic clades, we found that disease and carriage isolates showed no phylogenetic clustering. SSTI pathogenicity exhibited high narrow-sense heritability. Surface-associated LPXTG-anchored proteins, particularly spsF, harbored the strongest associations, with additional signals in iron metabolism (narH, sufB) and oxidative stress tolerance (ahpC). Random Forest classification validated GWAS signals. SSTI pathogenicity in S. pseudintermedius reflects a polygenic architecture driven by cumulative variation across surface-associated, metabolic, and stress-response loci, shifting focus from single virulence genes to genome-wide genetic variation. Full article

Background: Polygenic risk scores (PRS) have proven valuable for disease risk prediction, but their predictive utility often remains limited because human traits result from complex interactions between environmental and genetic factors. Blood lipid levels are heritable and clinically important risk factors for cardiovascular disease, yet it remains unclear whether multi-omics integration can enhance lipid trait prediction beyond PRS alone. Methods: We first constructed single-omics scores, where gene expression, plasma protein, and plasma/serum metabolite levels were genetically predicted and weighted by effect sizes estimated via LASSO regression. Subsequently, we implemented two integration strategies to develop composite multi-omics risk scores (MoRS): step-MoRS, which integrates single-omics scores using stepwise regression, and Lasso-MoRS, which directly models all predicted features across omics layers using LASSO regression. Both approaches were evaluated across European, South Asian, and African ancestries within the UK Biobank. Results: MoRS-based methods consistently demonstrated superior predictive accuracy compared to PRS alone for four lipid traits across diverse populations. Notably, Lasso-MoRS prioritized key biomarkers with predictive utility complementary to genomic data. Conclusions: These findings confirm that integrating multi-omics biomarkers with genomic data significantly enhances lipid trait prediction across diverse ancestries, offering biological insights into the molecular regulation of lipid metabolism. Full article