Bioinformatics and machine learning approaches reveal key genes and underlying molecular mechanisms of atherosclerosis: A review

Abstract

Atherosclerosis (AS) causes thickening and hardening of the arterial wall due to accumulation of extracellular matrix, cholesterol, and cells. In this study, we used comprehensive bioinformatics tools and machine learning approaches to explore key genes and molecular network mechanisms underlying AS in multiple data sets. Next, we analyzed the correlation between AS and immune fine cell infiltration, and finally performed drug prediction for the disease. We downloaded GSE20129 and GSE90074 datasets from the Gene expression Omnibus database, then employed the Cell-type Identification By Estimating Relative Subsets Of RNA Transcripts algorithm to analyze 22 immune cells. To enrich for functional characteristics, the black module correlated most strongly with T cells was screened with weighted gene co-expression networks analysis. Functional enrichment analysis revealed that the genes were mainly enriched in cell adhesion and T-cell-related pathways, as well as NF-κ B signaling. We employed the Lasso regression and random forest algorithms to screen out 5 intersection genes (CCDC106, RASL11A, RIC3, SPON1, and TMEM144). Pathway analysis in gene set variation analysis and gene set enrichment analysis revealed that the key genes were mainly enriched in inflammation, and immunity, among others. The selected key genes were analyzed by single-cell RNA sequencing technology. We also analyzed differential expression between these 5 key genes and those involved in iron death. We found that ferroptosis genes ACSL4, CBS, FTH1 and TFRC were differentially expressed between AS and the control groups, RIC3 and FTH1 were significantly negatively correlated, whereas SPON1 and VDAC3 were significantly positively correlated. Finally, we used the Connectivity Map database for drug prediction. These results provide new insights into AS genetic regulation.