Multi-Omics Integration Reveals Key Genes and Potential Metabolic Mediation Mechanisms in Atopic Dermatitis.
文献摘要
Objective To integrate expression quantitative trait locus-based Mendelian randomization (eQTL-MR), spatial transcriptomics, summary-data-based Mendelian randomization (SMR), mediation MR, and exploratory virtual knockout to prioritize candidate genes and potential metabolic mediation pathways in atopic dermatitis (AD). Methods Whole-blood eQTL and AD genome-wide association study data were analyzed by two-sample MR, followed by intersection with differentially expressed genes from lesional versus non-lesional AD skin spatial transcriptomics, SMR, and the heterogeneity in dependent instruments (HEIDI) test. MR of 1,400 plasma metabolomic traits was integrated with gene-metabolite and gene-AD estimates. Spatial expression and scTenifoldKnk applied to spot-level expression matrices were used for exploratory functional interpretation. Results MR identified 1,671 candidate genes; 66 overlapped with spatial transcriptomic differentially expressed genes, and 13 were prioritized by SMR/HEIDI. Eighty-six metabolites and 198 of 1,118 gene-metabolite tests were nominally associated (IVW P Conclusion These computational findings suggest candidate gene-metabolite relationships in AD. The mediation paths did not survive correction for multiple testing. Interpretation of the spatial expression and virtual-knockout results requires additional methodological validation and independent experimental confirmation.