Identification of signature of gene expression in biliary atresia using weighted gene co-expression network analysis. Issue 37 (16th September 2022)

• Record Type: Journal Article
• Authors: Wang, Yongliang; Yuan, Hongtao; Zhao, Maojun; Fang, Li
• Is Part Of: Medicine. Volume 101:Issue 37(2022)
• Page Start: e30232
• Publication Date: 2022-09-16
• Subjects: biliary atresia -- computational biology -- statistical model; Medicine -- Periodicals; Medicine -- Periodicals; Médecine -- Périodiques; Geneeskunde; Medicine; Periodicals; Periodicals; 610.5
• DOI: 10.1097/MD.0000000000030232
• Languages: English
• ISSNs: 0025-7974

Abstract

Abstract : Biliary atresia (BA) is the most common cause of obstructive jaundice during the neonatal period. This study aimed to identify gene expression signature in BA. The datasets were obtained from the Gene Expression Omnibus database. Weighted gene co-expression network analysis identified a critical module associated with BA, whereas Gene Ontology (GO) enrichment analysis and Kyoto Encyclopedia of Genes and Genomes pathway enrichment analysis revealed the functions of the essential modules. The high-connectivity genes in the most relevant module constructed protein–protein interaction networks via the string website and Cytoscape software. Hub genes screened by lasso regression consisted of a disease classification model using the randomforest method. Receiver operating characteristic curves were used to assess models' sensitivity and specificity and the model was verified using the internal and external validation sets. Ten gene modules were constructed by WGCNA, of which the brown module had a strong positive correlation with BA, comprising 443 genes. Functional enrichment analysis revealed that module genes were mainly involved in biological processes, such as extracellular matrix organization, cell adhesion, inflammatory response, and the Notch pathway ( P < .001), whereas these genes were involved in the metabolic pathways and cell adhesion molecules ( P < .001). Thirty-nine high-connectivity genes in the brown module constructed protein-protein interaction networks. keratin 7 ( KRT7 ) and C-X-C motif chemokine ligand 8 ( CXCL8 ) were used to construct a diagnostic model that had an accuracy of 93.6% and the area under the receiver operating curves for the model was 0.93. The study provided insight into the signature of gene expression and possible pathogenesis of BA; furthermore, it identified that the combination of KRT7 and CXCL8 could be a potential diagnostic model for BA.