Use this URL to cite or link to this record in EThOS:
Title: Using the Bayesian Normal Gamma prior to identify associated sequence variants
Author: Boggis, E. M.
ISNI:       0000 0004 5364 5021
Awarding Body: University of Sheffield
Current Institution: University of Sheffield
Date of Award: 2014
Availability of Full Text:
Access from EThOS:
Access from Institution:
The Normal Gamma prior, a Bayesian adaptive shrinkage method which is implemented using MCMC, is compared to other statistical methods as an eQTL approach to identifying causal or associated genetic mutations. The methods are compared on simulated data, where the results show the Normal Gamma prior to be a far superior method. On human data it is more difficult to assess the results for accuracy, but we can conclude that the Normal Gamma prior highlights SNPs in concordance with other methods. We also note that the Normal Gamma prior, although enforcing very harsh shrinkage, reports many less false positive SNPs than other methods. We develop the Normal Gamma prior to include functional information which we use to differentially penalise synonymous and nonsynonymous SNPs, as well as intronic, intergenic, splicing, UTR3 and other SNPs where necessary. In initial simulation studies, the prior distribution penalises synonymous SNPs on average more than non-synonymous SNPs. Further developments increase the penalisation on intronic, intergenic, UTR3, synonymous and other SNPs more than splicing and non-synonymous SNPs due to larger functional significance scores for the latter. The effect of this on the differential shrinkage between the two sets of SNPs can be seen in the posterior rankings and effect size estimates. We believe that this differential shrinkage form of the Normal Gamma prior is a powerful tool for detecting causal or associated SNPs, and has been shown to increase the posterior mean effect size estimates for causal SNPs with respect to the standard Normal Gamma, as well as increasing the ranking of validated causal SNPs (with respect to the standard Normal Gamma).
Supervisor: Walters, K. ; Milo, M. Sponsor: Not available
Qualification Name: Thesis (Ph.D.) Qualification Level: Doctoral
EThOS ID:  DOI: Not available