Scholarlink

<p>Multiple sclerosis (MS) is a complex autoimmune disease with strong genetic components, but its genetic mechanisms remain largely underexplored. We aimed to pinpoint causal genes and evaluate their utility for MS risk prediction. We integrated MS genome-wide association study summaries with brain-derived splicing quantitative trait loci (sQTLs) and expression quantitative trait loci (eQTLs) via summary-data-based Mendelian randomization (SMR) and colocalization analyses to identify potential causal genes. Weighted gene coexpression network analysis (WGCNA) of the E-MTAB-5151 dataset identified MS-associated gene modules. LASSO regression determined the core gene signature. GO and KEGG enrichment analyses, immune infiltration, and gene set enrichment analysis (GSEA) explored the biological relevance. Using an independent protein quantitative trait loci (pQTL) dataset, key genes were further validated for pQTL-MS associations. SMR identified 28 sQTL genes and 66 eQTL genes for MS, 23 and 51 of which passed the colocalization tests, respectively. WGCNA identified three MS-associated modules, and their intersection with SMR genes prioritized 23 key genes. Functional enrichment analysis of the module genes and SMR genes highlighted the consistent involvement of immune-related pathways in MS, including lymphocyte activation and NF-κB signalling. LASSO regression established a 10–gene signature (<i>ACP2</i>, <i>IL7</i>, <i>MYNN</i>, <i>RGS1</i>, <i>SAE1</i>, <i>SP140</i>, <i>TRAF3</i>, <i>TSPAN31</i>, <i>TYMP</i>, and <i>ZC2HC1A</i>) with high predictive accuracy (AUC = 0.983 in internal validation; AUC &gt; 0.70 across three external datasets). Immune infiltration analysis revealed a consistent immune cell expression pattern, in which the expression of MS risk genes was positively associated with naive CD4<sup>+</sup> T cells and resting mast cells, but negatively associated with activated mast cells. In contrast, MS protective genes exhibited the opposite pattern. Furthermore, the integration of the MS genome-wide association study statistics validated ZC2HC1A and TRAF3 at the protein level. GSEA further linked both genes to the Hedgehog signalling pathway. Integrating genomic, transcriptomic, and proteomic data, we identified candidate causal genes for MS with robust evidence. ZC2HC1A and TRAF3 have emerged as promising biomarkers and mechanistic candidates for MS. Future follow-up functional studies are warranted to elucidate their molecular roles in MS pathogenesis.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Integrated multi-omics and machine learning prioritize key immune genes for multiple sclerosis risk prediction

  • 期刊论文

  • 发表日期：

    2026年3月2日

查看全文
  • Ming Chen,
  • Duran Zhao,
  • Haiping Fan,
  • Xiaojun Zeng,
  • Wei Zhang,
  • Lijuan Li,
  • Wei Li

摘要

Multiple sclerosis (MS) is a complex autoimmune disease with strong genetic components, but its genetic mechanisms remain largely underexplored. We aimed to pinpoint causal genes and evaluate their utility for MS risk prediction. We integrated MS genome-wide association study summaries with brain-derived splicing quantitative trait loci (sQTLs) and expression quantitative trait loci (eQTLs) via summary-data-based Mendelian randomization (SMR) and colocalization analyses to identify potential causal genes. Weighted gene coexpression network analysis (WGCNA) of the E-MTAB-5151 dataset identified MS-associated gene modules. LASSO regression determined the core gene signature. GO and KEGG enrichment analyses, immune infiltration, and gene set enrichment analysis (GSEA) explored the biological relevance. Using an independent protein quantitative trait loci (pQTL) dataset, key genes were further validated for pQTL-MS associations. SMR identified 28 sQTL genes and 66 eQTL genes for MS, 23 and 51 of which passed the colocalization tests, respectively. WGCNA identified three MS-associated modules, and their intersection with SMR genes prioritized 23 key genes. Functional enrichment analysis of the module genes and SMR genes highlighted the consistent involvement of immune-related pathways in MS, including lymphocyte activation and NF-κB signalling. LASSO regression established a 10–gene signature (ACP2, IL7, MYNN, RGS1, SAE1, SP140, TRAF3, TSPAN31, TYMP, and ZC2HC1A) with high predictive accuracy (AUC = 0.983 in internal validation; AUC > 0.70 across three external datasets). Immune infiltration analysis revealed a consistent immune cell expression pattern, in which the expression of MS risk genes was positively associated with naive CD4+ T cells and resting mast cells, but negatively associated with activated mast cells. In contrast, MS protective genes exhibited the opposite pattern. Furthermore, the integration of the MS genome-wide association study statistics validated ZC2HC1A and TRAF3 at the protein level. GSEA further linked both genes to the Hedgehog signalling pathway. Integrating genomic, transcriptomic, and proteomic data, we identified candidate causal genes for MS with robust evidence. ZC2HC1A and TRAF3 have emerged as promising biomarkers and mechanistic candidates for MS. Future follow-up functional studies are warranted to elucidate their molecular roles in MS pathogenesis.