Integrated multi-omics profiling reveals shared mechanistic pathways in asthma-allergic rhinitis comorbidity: A hybrid machine learning framework leveraging Mendelian randomization for precision diagnostics.

Background: Asthma (AS) and allergic rhinitis (AR), though sharing Th2-driven inflammation, exhibit distinct clinical trajectories, with molecular mechanisms underlying their comorbidity remaining poorly characterized. This study aimed to delineate conserved and divergent immunometabolic pathways and develop a blood-based diagnostic framework integrating multi-omics biomarkers. Methods: We harmonized 8 peripheral blood transcriptomic cohorts (n = 1,073) using ComBat correction, performed Mendelian randomization (MR) across 5 asthma cohorts (FinnGen Release 12), and constructed a combinatorial machine learning model (113 configurations) validated in 5 independent cohorts. Results: Bidirectional regulation of the Y-chromosomal gene RPS4Y1 (upregulated in AR, downregulated in AS) was linked to disease-specific immunometabolic reprogramming (P < 0.001). MR identified 7 causal plasma proteins (GRAMD1C, GSTO1, IL1RAP, MMP9, PDXK, SAT2, SIGLEC12) intersecting transcriptomic signatures, implicating oxidative stress as a shared mechanism. The glmBoost-RF diagnostic model integrating these biomarkers achieved superior accuracy (AUC >95% in 4/5 cohorts), outperforming conventional classifiers reliant on IgE or eosinophils. Immune profiling revealed AR-specific native B-cell expansion and Treg depletion versus AS-associated CD4 Conclusions: This integrative analysis establishes RPS4Y1 as a central regulator of allergic inflammation dimorphism and delivers a validated multi-omics classifier for precision diagnostics. The findings bridge molecular sex differences, metabolic-immune crosstalk, and clinical heterogeneity, advancing phenotype-specific therapeutic strategies.