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

Autism spectrum disorder (ASD) is a neurodevelopmental disorder, caused by various epigenetic and genetic factors. This has resulted in an unclear understanding of etiology and biomarkers associated with ASD. In this study, we used RNA Seq datasets integrating with machine learning models to identify the differentially expressed genes (DEGs) between ASD and typical development (TD). The RNA Seq data of 20 ASD and 19 TD were obtained from the NCBI GEO database. We used standard R programming pipeline to pre-process the data. Further, random forest (RF) and eXtreme gradient boosting (XGBoost) models were employed to identify potential genes discriminate between ASD and TD. The model's performance was assessed using evaluation metrics and the 5-fold-cross validation. The proposed XGBoost models achieved a 5-fold cross validation accuracy, sensitivity, specificity, precision and F1-Score of 67.14%, 67.5%, 67.5%, 67.84%, 75.56% respectively. Further, TCTA gene was common in the top 10 gene signatures identified from RF and XGBoost. These results may aid in developing potential and prognostic biomarkers for effective ASD screening.