Using Haplotype Blocks for Detecting Interactions with Multifactor Dimensionality Reduction

• Activity Number: 188
• Type: Contributed
• Date/Time: Monday, August 1, 2011 : 10:30 AM to 12:20 PM
• Sponsor: ENAR
• Author(s): James Kniffen*+ and Nicole Mack and Alison Motsinger-Reif and David Reif
• Companies: North Carolina State University and North Carolina State University and North Carolina State University and North Carolina State University
• Keywords: statistical genetics ; haplotypes ; Multifactor Dimensionality Reduction ; gene-gene interactions ; linkage disequilibrium

Abstract

All genetic association analysis relies on linkage disequilibrium (LD), which describes the nonrandom assortment of genomic variants which is quantified as correlation between variants. This redundancy allows large sets of single-variant genetic data to be collapsed into haplotype blocks, which identify all correlated variant sites in a particular genomic region. Using haplotypes instead of single-variants can increase the power of gene-mapping studies due to denser data and reduction in number of tests. This is an important advantage as the scale of genetic studies creates challenges in variable selection, especially when searching for complex models including gene-gene interactions. In the current study we propose a haplotype collapsing approach to reduce the number of input variables for a commonly used approach for detecting interactions: Multifactor Dimensionality Reduction. We investigate this approach with simulated data to evaluate its performance and power to detect complex predictive models.

The research is based upon work supported by the National Science Foundation under CSUMS grant #DMS-0703392 (PI: Sujit Ghosh).