Abstract for presentation at 11th International Congress of Human Genetics

Linkage analysis using 19th century genealogical links inferred from genotypes

  • Jim Stankovich, Walter & Eliza Hall Institute of Medical Research, Australia
  • Dr Melanie Bahlo, Walter & Eliza Hall Institute of Medical Research, Australia
  • Dr Justin Rubio, Howard Florey Institute, Australia
  • Mr Stewart Huxtable, Walter & Eliza Hall Institute of Medical Research, Australia
  • Prof Trevor Kilpatrick, Howard Florey Institute, Australia
  • Prof Terence Speed, Walter & Eliza Hall Institute of Medical Research, Australia
  • Prof Simon Foote, Menzies Research Institute, Australia

    • Purpose: Genome-wide association studies have limited power to detect rare variants. There is more power to map variants by linkage analysis with multi-generational pedigrees; however such pedigrees can be difficult to identify. Our purpose is to develop a method for re-analysing data from genome-wide association studies, by identifying pedigrees from genotypes and then using linkage analysis.
    Methods: Previously we developed a hidden Markov model to infer 19th century genealogical links from genotypes. Applying the method to 170 subjects with multiple sclerosis (MS) from the Australian state of Tasmania genotyped with 800 microsatellites, we found 121 pairwise links connecting 104 of the cases (61%) with a false discovery rate of 10% (Hum Genet 117: 188 (2005)). In subsequent work Merlin has been used for linkage analysis with these inferred relationships. New methods were required to assess significance; this was done by permutation, repeating analyses with some case genotypes replaced by matched, untransmitted genotypes.
    Results: A region near chromosome 5cen has been identified where 7 distinct pairs of distantly-related cases have inherited haplotypes identical-by-descent (IBD) (p=0.0045). IBD sharing was confirmed by genotyping additional markers which were not used to infer relatedness. Suggestive linkage to MS in this region has been reported previously in the UK and Finland.
    Conclusions: While the power of this method will increase with denser marker sets, genealogical links will always be easier to find in founder populations such as Tasmania. Thus to maximise the power to map rare variants, we urge that some genome-wide association studies be conducted in founder populations. Unless the founder effect is severe, such populations are not a hindrance to fine-scale LD mapping of common variants; we have shown that LD patterns are very similar in Tasmania, another Australian state (Victoria), and the HapMap trios from Utah (Hum Genet Epub 11 Jan 2006).

    Conference Organiser - ICMS Pty Ltd