基于WGCNA-LASSO与机器学习建模的纤维肌痛综合征-骨质疏松症共病生物标志物研究OA

Objective To screen for comorbidity biomarkers of fibromyalgia syndrome(FMS)and osteoporosis(OP),and evaluate their diagnostic efficacy.Methods Datasets meeting the criteria for FMS and OP were screened from the Gene Expression Omnibus(GEO)database,and differentially expressed genes(DEGs)were identified in each dataset,respectively.In the FMS dataset,weighted gene co-expression network analysis(WGCNA)was performed to identify module genes highly correlated with FMS.The intersection of the above three gene sets was obtained using a Venn diagram to identify potential comorbidity genes,followed by Gene Ontology(GO)functional enrichment and Kyoto Encyclopedia of Genes and Genomes(KEGG)pathway enrichment analyses.LASSO regression analysis was then applied to the FMS and OP datasets to further screen the aforementioned potential comorbidity genes.After obtaining the core comorbidity genes,machine learning algorithms were employed to construct a diagnostic model to evaluate their diagnostic efficacy.Results By analyzing the FMS dataset,a total of 13 041 differentially expressed genes were screened;by analyzing the data from the OP dataset,a total of 410 differentially expressed genes were screened.After conducting WGCNA analysis on the FMS dataset,15 module genes were co-classified.After selecting four module genes with high correlation with the incidence of FMS and taking the intersection of differentially expressed genes,a total of 31 potential comorbid genes were obtained.The GO functional enrichment analysis results showed that it mainly participates in purine ribonucleoside metabolism and Th cell related biological processes;The KEGG pathway enrichment analysis results showed that it mainly participates in the chemical carcinogenic reactive oxygen species and oxidative phosphorylation signaling pathways.After further LASSO regression analysis of the 31 intersecting genes in the FMS dataset,15 genes could be obtained;After conducting LASSO regression analysis on the OP dataset,10 genes could be obtained.The intersection of the two resulted in four co pathogenic core genes,namely,proprotein convertase subtilisin/kexin Type 5(PCSK5),recombinant peptidylprolyl isomerase F(PPIF),recombinant pim-1 oncogene(PIM1),and zinc finger protein 528(ZNF528).Using four core comorbidity genes,diagnostic models for FMS and OP were constructed based on three machine learning algorithms,respectively,all with AUC values exceeding 0.7.The constructed nomogram model demonstrated accuracy and robustness,indicating its clinical utility.Conclusion The comorbidity mechanism of FMS and OP is complex,involving multiple biological processes and metabolic pathways.PCSK5,PPIF,PIM1 and ZNF528 play an important role in the occurrence and development of the comorbidity of FMS and OP.They have diagnostic value and can be used as biomarkers for the comorbidity of FMS and OP.