Classification autism spectrum disorder associated genes through analysis of the intrinsic disorder in their protein products

Autism Spectrum Disorder (ASD) affects approximately 1–2% of children worldwide, posing challenges for healthcare, education, and families. While genetic factors play a key role, only a small subset of ASD-related genes has been identified with strong genetic evidence. This study introduces a machine learning-based approach to enhance the classification of ASD-associated genes by analyzing the intrinsically disordered regions (IDRs) of their protein products. Using genetic data from the SFARI Gene database, 238 ASD-associated genes and 225 control genes were analyzed. Machine learning models, including artificial neural networks, were applied to classify ASD-related proteins based on IDR content and genetic patterns. The model achieved 66% accuracy and an AUC of 0.73, successfully distinguishing ASD-associated genes from non-ASD genes. Functional enrichment analysis revealed that proteins with high IDR content (≥ 49%) were linked to behavior and cognition, while proteins with low IDR content (≤ 11%) were associated with memory function. These findings suggest that IDRs play a crucial role in the functional specialization of ASD-associated proteins. This research demonstrates the potential of machine learning in ASD gene classification and provides insights into the biological mechanisms underlying ASD. Future studies could leverage these findings to improve genetic screening tools, diagnostics, and targeted therapies for ASD.