Integrative bioinformatics-machine learning-experimental validation identifies shared immunomodulatory targets in psoriasis and psoriatic arthritis and deciphers allicin's therapeutic mechanism.

Immune dysregulation is critically involved in psoriasis (PSO) and psoriatic arthritis (PsA). This study aims to identify shared immunomodulatory targets, predict potential therapeutics, and validate drug mechanisms for these diseases. Transcriptomic datasets were retrieved from GEO. Differential expression analysis was conducted using Limma. Weighted gene co-expression network analysis (WGCNA) and protein-protein interaction (PPI) networks were employed to identify key modules. Machine learning algorithms (Random Forest and LASSO) were applied to pinpoint critical immune-related genes. Diagnostic utility was assessed via ROC curves. Immune infiltration analysis and single-cell RNA sequencing (scRNA-seq) were used to evaluate immune dysregulation and cell-type-specific gene expression. Reverse drug screening was performed using Coremine Medical. Molecular docking and dynamics simulations assessed binding stability between candidate drugs and target proteins. RT-qPCR was utilized to examine drug-induced changes in gene expression and pathway activity. CXCL10, ISG15, and IFI27 were identified as hub genes. scRNA-seq confirmed their specific enrichment in immune cells. Garlic (Allium sativum) was prioritized through reverse screening, with allicin selected as its primary immunomodulatory component. Molecular simulations demonstrated stable binding of allicin to each target. Experimentally, allicin significantly upregulated ISG15 and downregulated CXCL10 and IFI27, likely via suppression of NF-κB signaling. CXCL10, ISG15, and IFI27 represent key immunotherapeutic targets in PSO and PsA. Allicin modulates these targets through NF-κB inhibition, demonstrating its potential as a therapeutic agent for autoimmune conditions.