Multi-omics integrated with machine learning identifies LPS-related genes potentially associated with iron metabolism-Immune axis imbalance in PCOS: a bioinformatics-based exploration of mechanisms and diagnostic markers

Abstract Chronic low-grade inflammation induced by bacterial lipopolysaccharide has been implicated in the pathogenesis of polycystic ovary syndrome; however, the genetic mechanisms linking lipopolysaccharide signaling to immune and metabolic dysregulation remain insufficiently elucidated. In the present study, transcriptomic datasets and single-cell sequencing data related to polycystic ovary syndrome were analyzed in combination with lipopolysaccharide-related genes retrieved from a toxicogenomics database. Differential expression analysis, weighted gene co-expression network analysis, clustering analysis, and machine learning algorithms were integrated to identify candidate biomarkers. Subsequently, functional enrichment analysis, immune cell infiltration analysis, regulatory network construction, and drug prediction analyses were conducted, while single-cell sequencing analysis was employed to identify key cellular populations and characterize gene expression dynamics. Two genes, C11orf68 and EVI5L, were identified as potential biomarkers and were significantly downregulated in patients with polycystic ovary syndrome. Functional analyses associated these genes with iron metabolism and immune regulation, whereas immune infiltration profiling identified T lymphocytes as key effector cells involved in disease progression. These findings suggest a potentially previously unrecognized association among lipopolysaccharide-related genes, iron metabolism imbalance, and immune dysregulation in polycystic ovary syndrome, thereby providing a potential framework for future mechanistic investigations and the development of diagnostic and therapeutic targets.