A hybrid framework for genome wide epistasis discovery

Area of Science:

• Genetics and Genomics
• Computational Biology
• Ophthalmology

Background:

• Genome-wide epistasis studies are crucial for understanding complex genetic traits.
• Identifying gene-gene interactions requires robust and efficient analytical frameworks.
• Vertical Cup-to-Disc ratio (vCDR) is a key indicator of optic nerve degeneration.

Purpose of the Study:

• To develop and validate a hybrid computational framework for large-scale genome-wide epistasis studies.
• To identify significant pairwise epistatic interactions associated with optic nerve degeneration (vCDR).
• To explore potential biological pathways implicated by identified genetic interactions.

Main Methods:

• A two-stage hybrid approach combining Random Forest and Logistic Regression models.
• Random Forest was used to pre-select potential epistasis-prone single nucleotide polymorphisms (SNPs).
• Logistic Regression was applied to identify significant pairwise epistatic SNP interactions.

Main Results:

• The framework identified 230 pairs of interacting SNPs with P-values < 5 × 10(-8).
• A high-confidence protein interaction network (score 0.9) was identified.
• Proteins involved in the WNT pathway, crucial for retinal ganglion cell survival, were implicated.

Conclusions:

• The proposed hybrid framework is effective and efficient for genome-wide epistasis studies.
• The identified genetic interactions and associated WNT pathway suggest a biological link to vCDR.
• This approach facilitates the discovery of complex genetic underpinnings of ocular diseases.