Identification of Diagnostic Markers for Major Depressive Disorder Using Machine Learning Methods

Area of Science:

• Genomics
• Bioinformatics
• Computational Biology

Background:

• Major Depressive Disorder (MDD) presents a significant global health burden, complicated by diverse symptom presentations that challenge clinical diagnosis.
• The need for robust biomarkers is critical for improving MDD diagnosis and understanding its underlying causes.
• Peripheral blood transcriptomes offer a potential source for developing diagnostic biomarkers for MDD.

Purpose of the Study:

• To develop a predictive model for diagnosing Major Depressive Disorder (MDD) using peripheral blood transcriptome data.
• To identify potential gene expression biomarkers associated with MDD.
• To evaluate the efficacy of machine learning algorithms for MDD classification.

Main Methods:

• Meta-analysis of nine RNA expression datasets from the Gene Expression Omnibus (GEO) database, including 302 samples from MDD patients and healthy controls.
• Application of the R package "MetaOmics" for genome-wide expression data analysis.
• Utilized machine learning algorithms including Support Vector Machine (SVM), Random Forest (RF), k-Nearest Neighbors (kNN), and Naive Bayesian (NB) for model construction and feature selection, with Receiver Operating Characteristic (ROC) curve analysis to assess diagnostic performance.

Main Results:

• Identified six differentially expressed genes (AKR1C3, ARG1, KLRB1, MAFG, TPST1, and WWC3) with a false discovery rate (FDR) < 0.05 between MDD patients and controls.
• Individual gene analysis showed moderate diagnostic ability (AUC range: 0.62-0.70).
• Machine learning models demonstrated superior diagnostic performance, with the SVM classifier achieving an AUC of 0.78 in an independent dataset, outperforming other models.

Conclusions:

• Meta-analysis of GEO data provides valuable insights into potential biomarkers for MDD.
• Machine learning models constructed using identified gene expression biomarkers offer a promising avenue for improving clinical diagnosis of MDD.
• The developed SVM classifier shows potential for practical application in distinguishing MDD from healthy controls.