Dear all,
I am pleased to announce a half-day meeting on `Statistical genetics' to
be held on Wednesday, 17th May, at the University of Leeds.
Further details appear below.
All welcome!
Regards, Paul
===================================================================
Dr. Paul D. Baxter
Secretary/Treasurer, RSS Leeds/Bradford Local Group,
Department of Statistics, University of Leeds, Leeds, LS2 9JT, UK.
-------------------------------------------------------------------
Leeds/Bradford: Wednesday 17th May, 3pm, Room 10.70, E.C. Stoner Building, Leeds University
(Refreshments from 2.30pm in the School of Mathematics)
Jenny Barrett (Division of Genetic Epidemiology, University of Leeds)
Statistical Challenges in the Design and Analysis of Whole Genome Association Studies
One of the main strategies in the search for genes that influence the risk of disease has been to
compare the distribution of genetic variants between cases with the disease and population-based
controls (genetic association studies). It is now becoming technically feasible to search the whole
genome for associated variants instead of targeting promising candidate genes. Preparations are
underway to test for disease associations with several hundred thousand genetic variants (single
nucleotide polymorphisms (SNPs)) spanning the genome. Statistical challenges that will be discussed
include the choice of SNPs to best capture variation in the genome, two-stage study designs, the
analysis of multiple SNPs and interpretation of results in the context of multiple testing. These
issues will be discussed in relation to a whole genome association study of melanoma skin cancer
currently underway in our group.
Mark Iles (Division of Genetic Epidemiology, University of Leeds)
Capturing Genetic Diversity in Association Studies
Ever denser maps of genetic markers (SNPs) are becoming available for genetic studies, allowing the
comprehensive evaluation of particular regions for disease susceptibility. Densely spaced SNPs are
likely to be strongly correlated, leading to much redundancy. Thus it is possible to capture most,
if not all, of the genetic variation in a region using only a subset of the available SNPs, known as
tagging SNPs. This can lead to a financial saving by reducing the amount of genotyping required and
perhaps make statistical analysis simpler.
However, the quality of the tagging SNPs selected is dependent on the initial sample in which they
are characterized. If the initial marker set is too sparse the tagging SNPs chosen will capture less
information than a naïve analysis suggests. A simple method has been proposed that should provide a
better estimate of the performance of tSNPs. We have investigated the magnitude of the bias using a
real data set and demonstrate, through simulation, that the novel method is both unbiased and
accurate, even for small numbers of typed markers
|