Whole-genome deep-learning analysis identifies contribution of noncoding mutations to autism risk

Area of Science:

• Genomics
• Computational Biology
• Neuroscience

Background:

• Noncoding mutations are challenging to link to complex diseases like autism spectrum disorder (ASD).
• Understanding the regulatory impact of genetic variants is crucial for disease gene discovery.

Purpose of the Study:

• To develop and apply a deep-learning framework for predicting the regulatory effects and deleterious impact of noncoding genetic variants.
• To investigate the role of noncoding mutations in the etiology of autism spectrum disorder (ASD).

Main Methods:

• A deep-learning framework was developed to predict regulatory effects and functional impact of genetic variants.
• The framework was applied to 1,790 autism spectrum disorder (ASD) simplex families.
• Analysis focused on de novo mutations in probands versus unaffected siblings.

Main Results:

• Noncoding mutations in ASD probands have a significantly higher functional impact than in unaffected siblings, affecting both transcriptional and post-transcriptional regulation.
• Noncoding mutations are implicated in synaptic transmission and neuronal development pathways in ASD.
• A link was identified between noncoding mutations and IQ heterogeneity in ASD.

Conclusions:

• Noncoding mutations play a significant role in autism spectrum disorder (ASD) pathogenesis.
• The predictive genomics framework is effective for identifying high-impact mutations and is applicable to other complex diseases.
• A convergent genetic landscape of both coding and noncoding mutations is emerging for ASD.