WGCNA combined with machine learning for analysis of diagnostic markers of preeclampsia associated with the hedgehog signaling pathway

Abstract

Background: Abnormal hedgehog (Hh) signaling is linked to preeclampsia (PE). This study aimed to identify Hh-related diagnostic biomarkers for PE and assess the role of immune infiltration. Methods: The PE dataset was obtained from GEO to screen DEGs. WGCNA was utilized to identify Hh pathway-related genes. Following the intersection of the two genes, key genes were screened by using LASSO, SVM-RFE, and RF. A model was constructed, with ROC applied for evaluating its performance. The ssGSEA was employed to analyze immune infiltration. Network Analyst was utilized to predict miRNA/TF targets. Results: Six Hh-related diagnostic genes were identified (SLC20A1, GPT2, PDK4, COASY, KRT81, CD163L1). The diagnostic model showed high accuracy (AUC > 0.8) in training and validation sets. PE patients exhibited immune dysfunction, including reduced dendritic cell, macrophage, and mast cell activity. Diagnostic genes strongly correlated with immune cells. Additionally, 25 miRNAs and 34 TFs potentially regulating these genes were predicted. Conclusions: Six potential PE diagnostic biomarkers were identified, and their immune interactions were explored. This study enhances understanding of PE pathogenesis and suggests therapeutic targets.