My apologies for the very late notice, but the following opportunity is available to researchers with at least 3 years postdoctoral experience:
"Translating genomewide association hits into functional candidates: a decision theory approach"
An abstract is given below. The post will be with Dr Michael Weale (Statistical Genetics Unit, KCL) and Prof Stephen Sacks (director of the MRC Centre for Transplantation, KCL), and the deadline for submission of applications in 5pm on Friday June 6th. Further details can be found here: http://www.gstt.nhs.uk/healthprof/researchanddevelopment/biomedicalresearch/careers/currentopps/trainingposts.aspx
Abstract of research project
The benefits of genomewide association studies (GWAs) have recently been demonstrated by a string of new discoveries of single nucleotide polymorphisms (SNPs) and Copy Number Variants that associate with complex phenotypes. The process of translating these discoveries into improved clinical practice is only just beginning, however. The MRC Centre for Transplantation at KCL has an ongoing project to define the genetic variants in a GWA scan of donor and recipient of renal allografts that contribute to progression to end stage renal failure and acute and chronic rejection, and a parallel project proposal has been developed to take forwards any hits for functional immunological follow-up.
In some instances, the process of nominating genes and candidate genetic variants for functional follow-up is relatively straightforward. This occurs when the signal for association is tightly localized around a single SNP, and the SNP is unambiguously sited in a single known gene. Often, however, the process of proceeding from associated SNPs to candidate functional genes is considerably complicated by the presence of more than one SNP with a strong association signal and by the presence of more than one candidate gene in the region. Furthermore, once the "low hanging fruit" of easily-defined instances have been taken forwards, the task of deriving as much as possible from the remaining association signals will acquire added importance as the only way to achieve maximum translational benefit from genomewide association methodology.
The purpose of this project will be to implement a decision theoretic framework for nominating candidate genes and candidate causal variants for functional follow-up. This Bayesian approach will be built upon the following themes:
1.. Inference of causal polymorphisms. Some recent methodological work by Weale and by others has provided a framework for assessing the posterior odds for causality of a set of SNPs in a candidate region, including SNPs that were not typed in the original GWA. This is done by combining GWA data with data from the HapMap genotyping resource.
2.. Inference of candidate functional genes. The framework outlined in Step 1 will be extended in this project. The question being asked will be inverted to become gene-based rather than SNP-based. An additional reversible jump Markov chain Monte Carlo step will add the true number of causal variants in the region as an additional parameter of interest. In addition, prior weights will be assigned to SNPs based on their position within or near putative functional genes. This will then provide a framework for assessing the posterior odds of causality of a set of genes in a candidate region.
3.. Decision theory for functional follow-up. The final step will incorporate the posterior odds of causality from Steps 1 and 2 and combine these with estimates of the costs and benefits that might be derived from functional follow-up, to achieve a formal decision theoretic model. We fully accept that such cost/benefits estimates are themselves subject to wide uncertainty themselves, and this uncertainty will be incorporated into the model. The eventual aim will be a tool that will guide rather than coerce the researcher in their choice of targets for functional follow-up. Both the scale of the task and the size of the datasets are too large to be tackled without computational help once the "low-hanging fruit" of easy targets have been exhausted. This is therefore a problem that would strongly benefit from a formal decision theoretic approach.
|