Genome-wide gene-gene interaction analysis for next-generation sequencing

Area of Science:

• Genetics
• Bioinformatics
• Statistical genomics

Background:

• Traditional interaction analysis for next-generation sequencing (NGS) data struggles with rare variants due to computational demands and low power.
• Challenges include paradigm shifts, severe multiple testing, and heavy computations in detecting genetic interactions.

Purpose of the Study:

• To develop a novel statistical method for interaction analysis in NGS data that overcomes limitations with rare variants.
• To shift the analysis paradigm from single nucleotide polymorphism (SNP) pairs to genomic regions for improved interaction detection.

Main Methods:

• Developed a novel statistic using functional data analysis techniques for dimensional reduction.
• Implemented functional logistic regression to collectively test interactions between all SNP pairs within two genomic regions.
• Applied the method to coronary artery disease (CAD) datasets (WTCCC, FHS) and early-onset myocardial infarction (EOMI) exome data.

Main Results:

• Functional logistic regression demonstrated correct type 1 error rates and higher power compared to existing methods in simulations.
• Identified 6 significantly interacted gene pairs in the Framingham Heart Study (FHS) replicated in the Wellcome Trust Case Control Consortium (WTCCC) study.
• Discovered 24 significantly interacted gene pairs in the early-onset myocardial infarction (EOMI) study after Bonferroni correction.

Conclusions:

• The proposed gene-region-based interaction analysis method is effective for NGS data, particularly for rare variants.
• The method offers improved power and computational efficiency over traditional SNP-pair analysis.
• The findings highlight significant gene interactions relevant to cardiovascular diseases, with potential for replication in independent cohorts.