Detecting genetic epistasis by differential departure from independence

Area of Science:

• Genomics
• Statistical Genetics
• Bioinformatics

Background:

• Epistasis (gene-gene interaction) is increasingly recognized as common, challenging prior assumptions of rarity.
• Current genome-wide association study (GWAS) practices often filter out genomic loci with low variant counts, potentially discarding significant epistatic signals.
• This filtering approach is suboptimal as it may overlook crucial epistatic patterns, especially those involving variants with nonuniform frequencies.

Purpose of the Study:

• To develop and validate a novel statistical test for inferring genetic epistasis in GWAS.
• To address the limitations of current methods that filter low-variant-count loci.
• To provide a robust method for detecting epistasis that is resilient to nonuniform variant frequencies.

Main Methods:

• Introduction of the compensated Sharma-Song test, which infers genetic epistasis by measuring differential departure from independence.
• Development of algorithms to simulate epistatic patterns exhibiting differential departures from independence.
• Comparative performance analysis of the compensated Sharma-Song test against the original Sharma-Song test and other alternatives using simulated data and a chicken abdominal fat content GWAS dataset.

Main Results:

• The compensated Sharma-Song test performed comparably to the original Sharma-Song test under marginally uniform variant frequencies.
• The test demonstrated a marked advantage over alternative methods when variant frequencies were marginally nonuniform.
• Application to chicken abdominal fat content data identified unique epistatic variants not prioritized by other methods, with associated genes belonging to obesity regulation pathways.

Conclusions:

• The compensated Sharma-Song test offers a practical and effective approach for studying epistasis in GWAS, particularly robust to nonuniform genetic variant frequencies.
• This method overcomes the limitations of filtering low-variant-count loci, enabling the detection of potentially missed epistatic interactions.
• The findings highlight the importance of considering epistasis in genetic studies and provide a valuable tool for uncovering complex genetic architectures.