Discovering pathway biomarkers of hepatocellular carcinoma occurrence and development by dynamic network entropy analysis

Chen Shen,Yi Cao,Jian Huang,Zhi-Ping Liu,Guo-qiang Qi

DOI: https://doi.org/10.1016/j.gene.2023.147467

IF: 3.913

2023-05-10

Gene

Abstract:Objective Gene expression profiling techniques measure the transcription of thousands of genes in a parallel manner. With more and more hepatocellular carcinoma (HCC) transcriptomic data becoming available, the high-throughput data provides an unprecedented opportunity to discover HCC diagnostic biomarkers. In this work, we propose a bioinformatics method based on dynamic network entropy analysis, called DNEA, to identify potential pathway biomarkers for HCC occurrence and development by integrating transcriptome and interactome. Methods We firstly collect the pathways documented in different knowledge-bases and then impose the genome-wide human transcriptomic data of multistage cancerous tissues during the development and progression of HCC. After linking the gene sets of pathways into individual connected networks, we map the corresponding gene expression information onto these pathways. The dynamic network entropy of individual pathways is calculated to evaluate its activities and dysfunctionalities during the disease occurrence and development. We use the overall significant difference in the entropic dynamics during the time course to prioritize distinctive pathways during disease progression. Then machine learning classification methods are employed to screen out pathway biomarkers with the classification ability to distinguish different-stage samples of HCC progression. Results Pathway biomarkers discovered based on DNEA demonstrate good classification performance in measuring HCC progression. The classification accuracy is as follows: DNA replication pathway (mean AUC= 0.82, 20 genes) from KEGG, FMLP pathway (mean AUC=0.84, 14 genes) from BioCarta, and downstream signaling of activated FGFR pathway (mean AUC =0.80, 15 genes) from Reactome. At the same time, previous studies have shown that these genes and pathways screened are closely related to the occurrence and development of HCC in terms of oncogenesis dysfunctions. Conclusions Our method for cancer biomarker discovery based on dynamic network entropy analysis is effective and efficient in identifying pathway biomarkers related to the progression of complex diseases.

genetics & heredity