Integrative Transcriptomic and Network Pharmacology Analysis Reveals Candidate Molecular Associations and Hypothesis-Generating Drug-Gene Relationships in Major Depressive Disorder

Abstract

Aims: To characterize transcriptomic alterations associated with major depressive disorder (MDD) in postmortem human brain tissue and integrate differential gene-expression findings with protein-protein interaction, functional-enrichment, and drug-gene interaction analyses. Study Design: Exploratory in silico transcriptomic and network pharmacology study based on secondary analysis of a publicly available RNA-sequencing dataset. Place and Duration of Study: Data were obtained from the NCBI Gene Expression Omnibus (GSE101521). Analyses were conducted using R and publicly available molecular-interaction and functional-annotation resources. Methodology: Postmortem dorsolateral prefrontal cortex RNA-sequencing data from 29 controls, 9 MDD non-suicide cases (MDD-NS), and 21 MDD suicide cases (MDD-S) were analyzed with DESeq2. A likelihood-ratio test (LRT) evaluated global three-group differences. A separate Wald contrast characterized CON-versus-MDD-NS differences. Seven prespecified transcriptomic candidates were carried forward to STRING and DGIdb analyses. Formal enrichment statistics were calculated for the expanded STRING network and interpreted separately from the original transcriptomic seeds. Results: The LRT identified HSPA6, IL1B, CCL4L2, CXCL8, SERPINH1, CCL2, and DMBT1L1 among the highest-ranked FDR-significant signals (adjusted P = 5.01 × 10⁻5 to 0.0316). In the exploratory CON-versus-MDD-NS contrast, HSPA6, IL1B, CXCL8, SERPINH1, and DMBT1L1 showed large positive log2FC estimates, indicating lower relative expression in MDD-NS; these pairwise findings are reported as nominal, unadjusted signals. The expanded STRING network comprised 16 nodes and 77 edges (12 expected; PPI enrichment P < 1.0 × 10⁻16). DGIdb returned 143 records, 131 (91.6%) of which involved CXCL8 or IL1B, indicating substantial target-specific annotation density. Conclusion: The data support exploratory molecular associations involving stress-response and immune/chemokine-related genes. Network expansion and database coverage materially influenced downstream enrichment and drug-gene results. The findings do not establish causal mechanisms, validated biomarkers, or therapeutic efficacy and require covariate-adjusted replication and experimental validation.