Machine learning identifies four key genes, DEFB4A, GJB2, SERPINB3, and SERPINB13, upregulated in psoriatic lesions

Psoriasis is a chronic, multifaceted inflammatory skin and systemic disorder driven by complex genetic, immunological, and environmental factors. While the IL-17/IL-23 immune axis is a central pathogenic pathway, a comprehensive understanding of all critical drivers, cytokines, and intracellular signaling networks remains incomplete. Current standard-of-care treatments often target broad inflammatory pathways, but identifying specific, novel therapeutic targets is crucial for developing more precise and effective interventions, addressing gaps in treatment efficacy and patient response variability.

Researchers analyzed single-cell RNA sequencing (scRNA-seq) datasets from both psoriatic and healthy skin samples, sourced from the Gene Expression Omnibus (GEO). Cell-type proportions were estimated using CIBERSORT, followed by weighted gene co-expression network analysis (WGCNA) to map correlations between specific cell types and gene signatures. Machine learning algorithms were then applied to refine the identified gene set, pinpointing DEFB4A, GJB2, SERPINB3, and SERPINB13 as key psoriasis-associated genes. Their expression was validated using bulk RNA-seq datasets, and their lesional regulatory roles and associated pathway alterations were further investigated to propose targeted therapeutic strategies.

A multi-algorithmic approach initially identified 271 hub genes significantly associated with psoriasis lesions and basal cells. Subsequent rigorous machine learning analysis successfully refined this extensive set, converging on a core group of four key driver genes implicated in psoriasis: DEFB4A, GJB2, SERPINB3, and SERPINB13. These four genes consistently demonstrated significant upregulation within lesional psoriatic skin compared to healthy controls. The study further explored the specific regulatory roles of these genes within the lesional microenvironment and their impact on associated cellular pathway alterations. While specific quantitative metrics like fold-changes or p-values for individual gene upregulation were not detailed in the abstract, the consistent identification and upregulation of these genes across analyses strongly suggest their pathogenic involvement. The researchers also screened small-molecule compounds that could potentially target these newly identified genes. > The machine learning analysis definitively identified DEFB4A, GJB2, SERPINB3, and SERPINB13 as key psoriasis-associated driver genes, all of which were found to be upregulated in lesional skin.