Abstract
AbstractMotivationSystemic lupus erythematosus (SLE) is an autoimmune disease and a long-term condition affecting many body parts. Autoimmune diseases are affecting more people for reasons unknown, and the causes of these diseases remain a mystery.MethodA newly introduced robust competing (risk) max-logistic regression classifier that can simultaneously perform subtype clustering and classification for disease diagnoses and predictions becomes a new hope to solve the mystery. We use this method in the study to discover critical DNA methylation CpG sites and genome genes, which lead to the highest accuracy and interpretability.ResultsThe DNA methylation CpG site cg05883128 (DDX60) and gene NR3C2 are essentially responsible for SLE development. They can lead to 100% prediction accuracy together with a miniature set of other CpG sites and genes, respectively. cg05883128 (DDX60) reveals the LSE mechanism affecting many body parts. NR3C2 in CD4 T cells and B cells behaves reversely, leading to the cause of LSE and explaining the mechanism of the autoimmune disease.ConclusionsThis work represents a pioneering effort and intellectual discovery in applying the max-logistic competing risk factor model to identify critical genes for LSE, and the interpretability and reproducibility of the results across diverse populations suggest that the CpGs and DEGs identified can provide a comprehensive description of the transcriptomic features of SLE. The practical implications of this research include the potential for personalized risk assessment, precision diagnosis, and tailored treatment plans for patients.
Publisher
Cold Spring Harbor Laboratory