Transcriptome Derived Artificial neural networks predict PRRC2A as a potent biomarker for epilepsy

Area of Science:

• Computational neuroscience
• Bioinformatics
• Genomics

Background:

• Epilepsy is a neurological disorder characterized by recurrent seizures due to abnormal brain electrical activity.
• Bioinformatics offers powerful tools for understanding neurological disorders at a molecular level and identifying diagnostic biomarkers.
• Artificial neural networks (ANNs) are computational models adept at analyzing complex biological datasets.

Purpose of the Study:

• To identify differentially expressed genes (DEGs) in epilepsy patients using bioinformatics approaches.
• To evaluate the potential of identified genes as diagnostic biomarkers for epilepsy.
• To develop and assess an artificial neural network model for epilepsy classification.

Main Methods:

• Retrieved and analyzed three NCBI GEO datasets (GSE190451, GSE140393, GSE134697) using the DESeq2 package for DEG identification.
• Employed WEKA software with various feature selection and classification algorithms to analyze DEGs.
• Constructed an Artificial Neural Network (ANN) model using R Studio with the identified significant genes and evaluated its performance using the 'pROC' R package.

Main Results:

• Identified seven up-regulated genes in epilepsy patients, with C4A later excluded due to low feature selection statistics.
• An ANN model utilizing six DEGs achieved an Area Under the Curve (AUC) of 0.720, indicating excellent classification accuracy.
• The gene PRRC2A demonstrated the highest generalized weight value in the ANN model, suggesting its significant role in epilepsy.

Conclusions:

• The study successfully identified PRRC2A as a potential diagnostic biomarker for epilepsy.
• Artificial neural networks combined with bioinformatics analysis provide a robust method for identifying epilepsy biomarkers.
• PRRC2A's elevated expression warrants further investigation for its clinical utility in epilepsy diagnosis.