Functional fine-mapping of noncoding risk variants in amyotrophic lateral sclerosis utilizing convolutional neural...

Area of Science:

• Genetics
• Neuroscience
• Artificial Intelligence

Background:

• Genome-wide association studies (GWAS) have identified common genetic variations linked to amyotrophic lateral sclerosis (ALS) risk.
• Identifying functional noncoding risk variants and their biological mechanisms in ALS remains a significant challenge.

Purpose of the Study:

• To develop a machine learning model to identify functional noncoding genetic variants associated with ALS using epigenetic data.
• To pinpoint novel risk variants and understand their impact on gene expression and transcription factor binding in ALS.

Main Methods:

• Constructed a convolutional neural network (CNN) model integrating large-scale GWAS meta-analysis data with epigenetic features.
• Filtered and prioritized candidate variants to fine-map specific risk loci.
• Analyzed the association of identified variants with gene expression levels and transcription factor binding sites.

Main Results:

• Identified two novel ALS risk variants, rs2370964 (chromosome 3) and rs3093720 (chromosome 17).
• These variants are associated with altered expression of CX3CR1 and TNFAIP1 genes.
• The polymorphisms affect transcription factor binding sites for CTCF, NFATc1, and NR3C1.

Conclusions:

• The study provides new insights into the pathogenesis of amyotrophic lateral sclerosis (ALS).
• The identified variants and their functional consequences offer potential targets for further research.
• The AI-driven methodology can be extended to investigate genetic underpinnings of other complex diseases.