Identification of cross-diagnostic biomarkers in ankylosing spondylitis and sarcopenia by bioinformatics and machine learning.

Ankylosing spondylitis (AS) and sarcopenia (SARC) often coexist, leading to impaired mobility through reduced muscle strength and altered bone metabolism. This study aimed to identify core diagnostic genes linking AS and SARC. This research analyzed 2 AS and 1 SARC dataset from the Gene Expression Omnibus database. Moreover, module genes and differentially expressed genes (DEGs) were evaluated via linear models for microarray data (Limma) and the weighted gene co-expression network analysis. Furthermore, functional enrichment analysis, various machine learning (ML) algorithms, and protein-protein interaction networks were employed for elucidating key candidate genes for the diagnosis of AS patients with SARC. The Receiver Operating Characteristic curve plots were utilized to determine the diagnostic significance of key genes. The merged AS dataset identified 1768 and 438 DEGs and module genes, respectively, in SARC. The intersection of module genes in SARC and DEGs in AS revealed 287 genes, which were predominantly enriched in oxidative phosphorylation. The protein-protein interaction network indicated 30 node genes. Furthermore, ML analysis identified 10 candidate hub genes for diagnostic value evaluation. In total, 6 candidate genes indicated high diagnostic significance key genes with the area under the curve > 0.7. The current study determined 6 hub genes (ENSA, FAM43A, MDH2, NUBP1, SAMM50, and TM2D1) for diagnosing AS patients with SARC, therefore providing a theoretical reference for potential diagnostic targets in these patients.