Integrative Multi-Omics and Mendelian Randomization Analysis Identify Core Genes and Potent Drug Targets for Insomnia

Insomnia is a common clinical sleep disorder, for which current medications only relieve symptoms symptomatically and highly specific therapeutic targets are limited. This study integrates multi-omics analysis and machine learning to screen and identify novel, safe, and efficient potential therapeutic targets for insomnia, thereby providing new insights for its targeted drug development and precision diagnosis and treatment.

We intersected a druggable gene set with blood, brain, and plasma Quantitative Trait Loci (QTL) data, using insomnia and its comorbidities as Genome-Wide Association Study (GWAS) outcomes. Two-sample Mendelian randomization (MR) analysis was used to assess causality, followed by sensitivity testing. The Gene Expression Omnibus (GEO) dataset GSE208668 was used for differential expression analysis and Weighted Gene Co-expression Network Analysis (WGCNA). Cross-validation and machine learning (LASSO regression and random forests) were employed to screen key hub genes and evaluate their diagnostic efficacy. Summary-data-based Mendelian Randomization (SMR), colocalization, and immune infiltration analyses were performed, and potential drugs were predicted via databases and molecular docking simulations.

Through multi-omics analyses including Mendelian randomization, transcriptome profiling and WGCNA, we identified 10 candidate genes, which were further refined by machine learning to five key hub genes: HLA-G, S1PR1, LGALS3, VIM, and PIK3CG. These genes were upregulated in insomnia patients and had high diagnostic AUC values. They showed extensive genetic associations with the risk of comorbidities such as depression, anxiety, hypertension, and diabetes. SMR and colocalization analyses provided convergent suggestive causal evidence for HLA-G, while the remaining four genes were supported as candidate targets rather than causal drivers. Immune infiltration analysis showed that key genes were associated with altered immune cell subsets. Drug prediction screened potential intervention compounds like decitabine and tamibarotene.

These findings provide robust evidence for novel druggable targets in insomnia, which exert regulatory effects via immune-inflammatory pathways and link insomnia to comorbidity risks. Their diagnostic potential and functional implications further underscore their translational value for clinical application and drug development, but all conclusions remain preliminary due to lack of independent validation.

This study identified five core genes, among which HLA-G shows suggestive causal links to insomnia, while the others are supported as promising candidate targets. They show promising diagnostic and drug development value, providing insights into insomnia pathogenesis and a theoretical basis for novel biomarkers and targeted therapies.