This paper aims to describe the benefits of using data recoding methods for the analysis of genetic interactions. By changing the representation of the input data it is possible to model non-additive genetic effects in association analysis software, which has been primarily designed to analyse only additive genetic effects. Similar treatment can be applied also for general-purpose statistical search algorithms available in general statistical packages. Data recoding is illustrated for several interaction models using hypothetical examples and by presenting gene-gene interaction analysis in a real cystic fibrosis dataset using the BAMA software.
Copyright 2008 S. Karger AG, Basel.