Gene-gene interactions detection using a two-stage model

Zhanyong Wang, Jae Hoon Sul, Sagi Snir, Jose A. Lozano, Eleazar Eskin

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review


Genome wide association studies (GWAS) have discovered numerous loci involved in genetic traits. Virtually all studies have reported associations between individual single nucleotide polymorphism (SNP) and traits. However, it is likely that complex traits are influenced by interaction of multiple SNPs. One approach to detect interactions of SNPs is the brute force approach which performs a pairwise association test between a trait and each pair of SNPs. The brute force approach is often computationally infeasible because of the large number of SNPs collected in current GWAS studies. We propose a two-stage model, Threshold-based Efficient Pairwise Association Approach (TEPAA), to reduce the number of tests needed while maintaining almost identical power to the brute force approach. In the first stage, our method performs the single marker test on all SNPs and selects a subset of SNPs that achieve a certain significance threshold. In the second stage, we perform a pairwise association test between traits and pairs of the SNPs selected from the first stage. The key insight of our approach is that we derive the joint distribution between the association statistics of a single SNP and the association statistics of pairs of SNPs. This joint distribution allows us to provide guarantees that the statistical power of our approach will closely approximate the brute force approach. We applied our approach to the Northern Finland Birth Cohort data and achieved 63 times speedup while maintaining 99% of the power of the brute force approach.

Original languageEnglish
Title of host publicationResearch in Computational Molecular Biology - 18th Annual International Conference, RECOMB 2014, Proceedings
PublisherSpringer Verlag
Number of pages16
ISBN (Print)9783319052687
StatePublished - 2014
Event18th Annual International Conference on Research in Computational Molecular Biology, RECOMB 2014 - Pittsburgh, PA, United States
Duration: 2 Apr 20145 Apr 2014

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume8394 LNBI
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349


Conference18th Annual International Conference on Research in Computational Molecular Biology, RECOMB 2014
Country/TerritoryUnited States
CityPittsburgh, PA

ASJC Scopus subject areas

  • Theoretical Computer Science
  • General Computer Science


Dive into the research topics of 'Gene-gene interactions detection using a two-stage model'. Together they form a unique fingerprint.

Cite this