Identification of Potential Biomarkers in ASD Integrating RNA Seq Data and Machine Learning Approaches

Area of Science:

• Genomics
• Neuroscience
• Computational Biology

Background:

• Autism spectrum disorder (ASD) is a complex neurodevelopmental condition with unclear etiology and limited biomarkers.
• Genetic and epigenetic factors contribute to ASD, necessitating advanced analytical approaches.

Purpose of the Study:

• To identify differentially expressed genes (DEGs) between individuals with ASD and typical development (TD) using RNA sequencing data.
• To leverage machine learning models for distinguishing ASD from TD based on gene expression profiles.

Main Methods:

• RNA sequencing datasets from 20 ASD and 19 TD individuals were analyzed.
• Standard R programming was used for data pre-processing.
• Random Forest (RF) and eXtreme gradient boosting (XGBoost) models were employed to identify key genes.

Main Results:

• XGBoost models achieved 67.14% accuracy, 67.5% sensitivity, 67.5% specificity, 67.84% precision, and 75.56% F1-Score via 5-fold cross-validation.
• The TCTA gene was identified as a common top gene signature by both RF and XGBoost models.
• These findings highlight potential diagnostic markers for ASD.

Conclusions:

• Machine learning analysis of RNA sequencing data can identify potential biomarkers for ASD.
• The TCTA gene shows promise as a discriminatory marker between ASD and TD.
• These results could contribute to the development of prognostic biomarkers for ASD screening.