Identification of lipid metabolism related immune markers in atherosclerosis through machine learning and experimental analysis

Background Atherosclerosis is a significant contributor to cardiovascular disease, and conventional diagnostic methods frequently fall short in the timely and accurate detection of early-stage atherosclerosis. Abnormal lipid metabolism plays a critical role in the development of atherosclerosis. Consequently, the identification of new diagnostic markers is essential for the precise diagnosis of this condition. Method The datasets related to atherosclerosis utilized in this research were obtained from the GEO database (GSE2470, GSE24495, GSE100927 and GSE43292). The ssGSEA technique was first utilized to assess lipid metabolism scores in samples affected by atherosclerosis, thereby aiding in the discovery of important regulatory genes linked to lipid metabolism via WGCNA. Following this, differential expression analysis and functional evaluations were carried out, after which various machine learning approaches were employed to determine significant diagnostic genes for atherosclerosis. A diagnostic model was then developed and validated through several machine learning algorithms. Furthermore, molecular docking studies were conducted to analyze the binding affinity of these key markers with therapeutic agents for atherosclerosis. The ssGSEA technique was also used to measure immune cell scores in atherosclerotic samples, aiding the exploration of the connection between key diagnostic markers and immune cells. Finally, the expression variations of the identified pivotal genes were confirmed through experimental validation. Result WGCNA identified 302 lipid metabolism-related genes in atherosclerotic samples, and functional analysis revealed that these genes are associated with multiple immune pathways. Through further differential analysis and screening using machine learning algorithms, APLNR, PCDH12, PODXL, SLC40A1, TM4SF18, and TNFRSF25 were identified as key diagnostic genes for atherosclerosis. The diagnostic model we constructed was confirmed to predict the occurrence of atherosclerosis with high accuracy, and molecular docking studies indicated that these six key diagnostic genes have potential as drug targets. Additionally, the ssGSEA algorithm further validated the association of these diagnostic genes with various immune cells. Finally, the expression levels of these six genes were experimentally confirmed. Conclusion Our study introduces novel lipid metabolism-related diagnostic markers for atherosclerosis and emphasizes their potential as immune-related drug targets. This research provides a valuable approach for the predictive diagnosis and targeted therapy of atherosclerosis.