Analysis of Potential Genomic Confounding in Genetic Association Studies and an Online Genomic Confounding Browser (GCB)

Authors

  • Christopher A. Raistrick,

    1. Bristol Genetic Epidemiology Laboratories, Department of Social Medicine, University of Bristol, Oakfield House, Oakfield Grove, Bristol, BS8 2BN, UK
    Search for more papers by this author
  • Khalid K. Alharbi,

    1. Clinical Laboratory Sciences Department, College of Applied Medical Sciences, King Saud University, P.O. Box 10219, Riyadh 11433, Saudi Arabia
    Search for more papers by this author
  • Ian N. M. Day,

    Corresponding author
    1. Bristol Genetic Epidemiology Laboratories, Department of Social Medicine, University of Bristol, Oakfield House, Oakfield Grove, Bristol, BS8 2BN, UK
    2. MRC Centre for Causal Analyses in Translational Epidemiology (CAiTE), Department of Social Medicine, University of Bristol, Oakfield House, Oakfield Grove, Bristol, BS8 2BN, UK
    Search for more papers by this author
  • Tom R. Gaunt

    1. Bristol Genetic Epidemiology Laboratories, Department of Social Medicine, University of Bristol, Oakfield House, Oakfield Grove, Bristol, BS8 2BN, UK
    2. MRC Centre for Causal Analyses in Translational Epidemiology (CAiTE), Department of Social Medicine, University of Bristol, Oakfield House, Oakfield Grove, Bristol, BS8 2BN, UK
    Search for more papers by this author

Ian N. M. Day, MRC Centre for Causal Analyses in Translational Epidemiology (CAiTE), Department of Social Medicine, University of Bristol, Oakfield House, Oakfield Grove, Bristol, BS8 2BN, UK. Tel: +44-(0)117-331-0098; Fax: +44 (0)117-331-0123; E-mail: ian.day@bristol.ac.uk

Summary

Genome-wide association studies have transformed genetic studies of disease susceptibility, identifying many variants that may tag functional polymorphism nearby. Variants are often ascribed to a physically close gene exhibiting plausible functionality for a causal pathway. However, more physically remote genes may be at a lesser linkage or linkage disequilibrium (LD) distance from the tested SNP and could therefore contain the functional variant tagged. This analysis aims to identify instances where research may be misled by misassociation of a variant with a gene and develop tools to analyse genomic confounding. A catalogue of reported associations was systematically analysed for unreported genes which may represent the true functionality ascribed to a reported variant, calculating physical and genetic distances for all genes within 1 cM of the tagging polymorphism. Results revealed 55 SNPs where recombination was lower between the identified SNP and a physically more remote gene than initially reported, and 374 where an alternative gene was genetically and physically closer than the reported gene. Analyses show potential for genomic confounding through false inferences of variant association to a gene. An online visualization tool (http://gcb.genes.org.uk/) was developed to plot genes by physical and genetic distance relative to a variant, along with LD data.

Ancillary