Mining phase separation-related diagnostic biomarkers for endometriosis through WGCNA and multiple machine learning techniques: a retrospective and nomogram study.

OBJECTIVE: The objective of this study was to investigate the role of phase separation-related genes in the development of endometriosis (EMs) and to identify potential characteristic genes associated with the condition.

METHODS: We used GEO database data, including 74 non-endometriosis and 74 varying-degree EMs patients. Our approach involved identifying significant gene modules, exploring gene intersections, identifying core genes, and screening for potential EMs biomarkers using weighted gene co-expression network analysis (WGCNA) and various machine learning approaches. We also performed gene set enrichment analysis (GSEA) to understand relevant pathways. This comprehensive approach helps investigate EMs genetics and potential biomarkers.

RESULTS: Nine genes were identified at the intersection, suggesting their involvement in EMs. GSEA linked DEGs to pathways like complement and coagulation cascades, DNA replication, chemokines, apical plasma membrane processes, and diseases such as Hepatitis B, Human T-cell leukemia virus 1 infection, and COVID-19. Five feature genes (FOS, CFD, CCNA1, CA4, CST1) were selected by machine learning for an effective EMs diagnostic nomogram. GSEA indicated their roles in mismatch repair, cell cycle regulation, complement and coagulation cascades, and IL-17 inflammation. Notable differences in immune cell proportions (CD4 T cells, CD8 T cells, DCs, macrophages) were observed between normal and disease groups, suggesting immune involvement.

CONCLUSIONS: This study suggests the potential involvement of phase separation-related genes in the pathogenesis of endometriosis (EMs) and identifies promising biomarkers for diagnosis. These findings have implications for further research and the development of new therapeutic strategies for EMs.