Transcriptomics and machine learning to advance schizophrenia genetics: A case-control study using post-mortem brain...

Area of Science:

• Neuroscience and Computational Biology
• Genomics and Bioinformatics

Background:

• Schizophrenia (SCZ) is associated with alterations in gene expression.
• Machine learning (ML) shows promise in analyzing gene expression data for SCZ research.

Purpose of the Study:

• To evaluate the performance of ML in classifying SCZ cases versus controls.
• To utilize gene expression microarray data from the dorsolateral prefrontal cortex for classification.

Main Methods:

• An XGBoost ML algorithm was trained and evaluated using 201 SCZ cases and 278 controls.
• 10-fold cross-validation and a held-out testing set were employed for model evaluation.
• Area Under the Receiver-Operator Characteristics Curve (AUC) was the primary performance metric.

Main Results:

• The ML model achieved an AUC of 0.76 on both cross-validation and testing data.
• An automated ML strategy improved classification performance, reaching an AUC of 0.79.
• Identified significant gene sets related to "oxidoreductase activity" and "integrin binding".

Conclusions:

• ML classification of SCZ based on gene expression data demonstrated above-chance performance.
• ML analysis of gene expression can enhance understanding of SCZ pathophysiology.
• This approach may facilitate the identification of novel therapeutic strategies for SCZ.