Multi-cohort transcriptomic analysis with machine learning identifies interferon-related candidate genes in systemic lupus erythematosus.

Systemic lupus erythematosus (SLE) is a heterogeneous autoimmune disease with complex molecular mechanisms. Although transcriptomic studies have revealed prominent interferon signatures in SLE, robust prioritization of candidate genes across independent cohorts remains challenging. In this study, we performed a multi-cohort transcriptomic analysis using publicly available GEO datasets. Differential expression analysis was conducted independently within each cohort to minimize cross-study confounding. Machine learning models, including LASSO, support vector machine, and random forest, were applied for feature prioritization in a designated training cohort, followed by independent validation in separate datasets. Model interpretability was assessed using SHAP analysis. Immune cell composition was estimated descriptively using CIBERSORT. In addition, molecular docking and molecular dynamics simulations were performed as exploratory in silico analyses to evaluate potential protein-compound interactions. A set of consistently dysregulated genes across cohorts was identified, many of which are associated with interferon signaling. Among these, RSAD2 showed robust prioritization across multiple machine learning models. The predictive performance of the models was stable in independent validation datasets. SHAP analysis highlighted the contribution of interferon-stimulated genes to model predictions. Immune deconvolution suggested altered immune cell composition in SLE samples, consistent with previously reported immune activation patterns. Exploratory in silico analyses suggested a potential interaction between artemisinin and RSAD2. This study provides a robust, multi-cohort computational framework for prioritizing candidate genes associated with SLE. The findings highlight interferon-associated transcriptional features as reproducible molecular signatures of SLE and generate testable hypotheses for future experimental and clinical investigation.