Integrative Bioinformatics and Multi-Method Machine Learning Reveal Robust Gene Signatures in a Mouse Model of PTSD

Posttraumatic stress disorder (PTSD) stems from severe trauma and is characterized by intrusive memories, avoidant behaviors, negative changes in cognition and mood, and hypervigilance. Current treatments, including selective serotonin reuptake inhibitors (SSRIs) and trauma-focused psychotherapy, are only partially effective for many patients, highlighting the need for a deeper mechanistic understanding. This study analyzed transcriptome data from a social stressor-induced PTSD mouse model (GSE45035), focusing on the amygdala and medial prefrontal cortex (mPFC) after 5 or 10 days of chronic social stress. Differential expression analysis showed that gene expression changes in the amygdala were significantly greater than in the mPFC, especially highlighting region- and time-specific responses. Gene ontology (GO) and KEGG pathway enrichment analysis showed that differentially expressed genes (DEGs) in the amygdala were mainly involved in synaptic transmission and plasticity, while those in the mPFC were associated with transcriptional regulation and metabolic processes. We identified a robust set of candidate genes associated with neuronal remodeling and adaptation to chronic stress by employing five complementary machine learning-based feature selection approaches. Altogether, these findings provide new insights into the pathophysiology of PTSD and offer potential molecular targets for more effective therapeutic strategies.