Construction of the gene expression subgroups of patients with coronary artery disease through bioinformatics approach

Coronary artery disease (CAD) is a heterogeneous disease that has placed a heavy burden on public health due to its considerable morbidity, mortality and high costs. Better understanding of the genetic drivers and gene expression clustering behind CAD will be helpful for the development of genetic diagnosis of CAD patients. The transcriptome of 352 CAD patients and 263 normal controls were obtained from the Gene Expression Omnibus (GEO) database. We performed a modified unsupervised machine learning algorithm to group CAD patients. The relationship between gene modules obtained through weighted gene co-expression network analysis (WGCNA) and clinical features was identified by the Pearson correlation analysis. The annotation of gene modules and subgroups was done by the gene ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway analysis. Three gene expression subgroups with the clustering score of greater than 0.75 were constructed. Subgroup I may experience coronary artery disease of an in-creased severity, while subgroup III is milder. Subgroup I was found to be closely related to the upregulation of the mitochondrial autophagy pathway, whereas the genes of subgroup II were shown to be related to the upregulation of the ribosome pathway. The high expression of APOE, NOS1 and NOS3 in the subgroup I suggested that the patients had more severe coronary artery disease. The construction of genetic subgroups of CAD patients has enabled clinicians to improve their understanding of CAD pathogenesis and provides potential tools for disease diagnosis, classification and assessment of prognosis.