The effect of genotype imputation on the validity and power of statistical tests for gene-environment interactions in case-only studies

Principal Investigator

Associated Principal Investigator

Background and current state of research

Gene-environment (GxE) interactions are involved in the etiology of many complex diseases such as chronic inflammatory bowel diseases. A powerful design for the statistical analysis of GxE interactions in epidemiological studies is the case-only (CO) design where only information on diseased cases, but not on healthy controls, is required. A large number of genetic markers are usually genotyped, but some genotypes may be missing, e.g. due to technical reasons. These can be imputed based on reference haplotypes of healthy individuals (“in-silico genotyping”). Several implementations of genotype imputation exist, but there is no thorough investigation yet of the impact of imputation on the validity and power of statistical association tests for GxE interaction in studies using a case-only design.

Our goals

The aims of this project are to

  • quantify the impact of imputation accuracy and choice of imputation basis on the validity and power of detecting GxE using a CO design, and
  • investigate GxE for inflammatory diseases on a genome-wide scale using imputed genotype data available from collaborators within the RTG.

How to get there

Research plan for the doctoral student:

Investigate the theoretical effect of imputation based on healthy controls in a CO study using population genetic models of LD structure. Validate imputation accuracy using simulations based on real Immunochip data, which provides the true LD structure in cases both in unrelated and in truly associated genetic regions. Investigate the validity and power of a GxE interaction test in a CO design using Immunochip data by simulating exposure status for specified GxE effects. Analysis of real data on GxE interactions from other projects of the RTG.