An explainable predictive machine learning model reveals ARRB2 as a key gene in post-traumatic stress disorder: A GEO...

Area of Science:

• Neuroscience and Genetics
• Computational Biology and Bioinformatics

Background:

• Post-Traumatic Stress Disorder (PTSD) development involves complex pathways, necessitating new diagnostic and management biomarkers.
• Current understanding of PTSD pathogenesis requires further molecular insights for improved clinical applications.

Purpose of the Study:

• To identify novel molecular biomarkers for Post-Traumatic Stress Disorder (PTSD) using integrated bioinformatics and machine learning approaches.
• To validate the identified biomarkers in relevant brain regions and assess their diagnostic potential.

Main Methods:

• Differential gene expression analysis using Limma, weighted gene co-expression network analysis (WGCNA), and five machine learning algorithms for gene screening.
• Development of a predictive model using eight machine learning algorithms, with Support Vector Machine (SVM) showing the highest efficiency (AUC=0.894).
• Feature importance analysis using Shapley Additive exPlanations (SHAP), immune-cell infiltration analysis, and experimental validation in rat models (SPS&S) and brain tissues (PFC, HIP, AMY).

Main Results:

• Five key genes were identified as potential PTSD biomarkers.
• The SVM model demonstrated high predictive efficiency for PTSD classification.
• Immune cell infiltration patterns differed between controls and PTSD patients, correlating with key genes. Arrestin beta 2 (ARRB2) gene expression was significantly downregulated in the hippocampus (HIP) of rats, confirmed by Western Blotting and immunofluorescence.

Conclusions:

• The study successfully identified Arrestin beta 2 (ARRB2) as a potential biomarker for Post-Traumatic Stress Disorder (PTSD).
• Integrated machine learning, bioinformatics, and experimental validation provide a robust framework for biomarker discovery in PTSD.
• ARRB2's downregulation in the hippocampus suggests a role in PTSD pathophysiology and potential diagnostic utility.