Multi-omics Data Integration with Consensus Clustering Ensemble for Lower-grade Gliomas Cancer Subtype Identification
10.11783/j.issn.1002-3674.2025.04.005
- VernacularTitle:低级别胶质瘤多组学数据整合的一致性聚类集成分子分型
- Author:
Tong WANG
1
;
Qi YANG
;
Yaxin TIAN
Author Information
1. 山西医科大学医学科学院 030001;山西医科大学公共卫生学院卫生统计教研室,重大疾病风险评估山西省重点实验室
- Publication Type:Journal Article
- Keywords:
Clustering ensemble;
Multi-omics data;
Molecular subtype;
Lower-grade gliomas
- From:
Chinese Journal of Health Statistics
2025;42(4):502-509
- CountryChina
- Language:Chinese
-
Abstract:
Objective To identify subtypes of lower-grade gliomas based on multi-omics data integration with consensus clustering ensemble(MICCE)method,and further assess prognosis risk across different subtypes and explore differentially expressed biomarkers and pathways.Methods We applied the consensus clustering ensemble method to integrate the subtype results of seven multi-omics data integration methods(SNF,joint SNF,CIMLR,ConsensusClusterPlus,MoCluster,NEMO,iClusterBayes)for mRNA,miRNA,and DNA methylation data from LGG patients,identifying a robust molecular subtyping.Then we performed survival analysis based on the subtype results,and Cox proportional risk models were fitted to assess the prognosis of patients with different subtypes.Differentially expressed genes(DEmiRNAs,DEmRNAs and DMGs)between different subtypes were screened,and GO(gene ontology)analysis and KEGG enrichment analysis were performed for overlapping genes among DEmiRNAs target genes,DEmRNAs,and DMGs.Ultimately,immune infiltration analysis and pathway activity analysis were conducted to quantify the biological differences among different subtypes.Results Patients were classified into three subtypes:a high-risk cluster,a moderate-risk cluster,and a low-risk cluster.The results showed that the high-risk cluster were 7.70 times more likely to die than patients in low-risk cluster.A total of 2512 DEmRNAs,14 DEmiRNAs and 255 DMGs were screened,the combined analysis genes yielded 665 genes which are regulated by mRNA,miRNA and DNA methylation and enriched 62 GO items and 52 KECG pathways with statistical differences.The analysis of immune infiltration and pathway activity indicates that there are two immune cells and four signaling pathways with statistically significant differences.Conclusion MICCE can effectively identify high-risk patients of LGG.Subsequent analysis reveals differential genes and pathways related to the progression of LGG with different subtypes,providing important clues for the personalized treatment of LGG.
