Discovering new interaction between genes in extrahepatic Cholangiocarcinoma using Bayesian networks

Extrahepatic Cholangiocarcinoma (eCCA) is one of the most severe cancers, owing to ineffective therapeutics and resistance to treatments. There has been a lack of knowledge on the pathogenesis of eCCA, which necessitates the investigation of fine biomarkers and the discovery of targeted therapies. Bayesian networks (BNs) are known as impactful tools for dissecting gene-gene interaction in complex biological networks. Initially, we identified differentially expressed genes in eCCA using the GSE132305 dataset from GEO. eCCA-related genes were obtained separately from DisGeNET. The overlap of these two gene sets resulted in a robust minimal gene list with known roles in eCCA. We then proceeded to select the highest up-and down-regulated genes to combine them with the overlapped genes. This led us to the final dataset which was later used as the input for "bnlearn" R package to generate BNs. Investigation of the generated BNs, led to the detection of candidate genes being connected to the parent nodes, namely APOC1, PIGR, FOSB, RPS6, FGA, CTSE, and ITIH2. Further literature review revealed that out of seven candidate genes, only two (FOSB and FGA) had been reported previously in the context of cholangiocarcinoma. The other five (APOC1, PIGR, RPS6, CTSE and ITIH2) have not been reported to have any roles in eCCA. Our results can potentially be the foundation for designing new studies on the mentioned genes to delve into the molecular mechanism behind the pathogenesis of eCCA and the discovery of novel drug targets. This study emphasizes the power of BNs combined with bioinformatics tools to unlock new avenues in the understanding and treatment of complex diseases such as eCCA.