Integrative transcriptomic and machine learning analyses identify NOL7 and TXLNA as candidate blood biomarkers for Parkinson’s disease within an RNA modification-associated co-expression network

Abstract Background Parkinson’s disease (PD) pathogenesis involves complex molecular mechanisms, with emerging evidence implicating RNA modifications (RM). This study sought to explore RM-associated key genes and their roles in PD progression. Methods Transcriptomic datasets GSE6613 (training) and GSE72267 (validation) were analyzed. Differentially expressed genes (DEGs) between PD and controls were recognized. WGCNA was performed to identify RM-associated module genes. Machine learning algorithms, Receiver Operating Characteristic (ROC) analysis, and gene expression validation were applied to screen key genes. Nomogram construction, functional enrichment, immune infiltration analysis were conducted to investigate the biological mechanisms and therapeutic potential of the key genes. The bioinformatics findings were further supported by RT-qPCR experiments in a small clinical cohort ( n = 5 per group), though these preliminary results require validation in larger independent samples. Results Through intersection of 507 DEGs and 2,091 RM-associated module genes, 63 candidate genes related to RM in PD were identified. Machine learning, ROC analysis, gene expression validation, and clinical experiments were further employed to identify two key genes (NOL7 and TXLNA). A nomogram constructed based on key genes demonstrated moderate diagnostic efficacy for PD, with an area under the curve of 0.792. Enrichment analyses revealed associations of the key genes with neuroactive pathways, such as spliceosome and ribosome. Immune infiltration analysis suggested a negative correlation between NOL7 and NKT (cor = -0.30, P < 0.05). Furthermore, danazol was predicted to be a compound associated with both NOL7 and TXLNA. Molecular docking analysis revealed that danazol exhibited a relatively favorable binding affinity for NOL7 (-6.3 kcal/mol), whereas its binding affinity for TXLNA was weaker (-4.7 kcal/mol). Conclusion NOL7 and TXLNA were validated as blood biomarkers for PD derived from an RNA modification-associated transcriptional module, offering insights into epigenetic dysregulation and immune interactions. The nomogram provided a preliminary framework for PD risk assessment, and the drug prediction offered potential candidates for future therapeutic exploration.