The goal of my thesis work was the statistical analysis of a genome-wide association study (GWAS) data from the patients suffering in ragweed pollen allergy and treated with allergen-specific immunotheraphy (ASIT). The main task was to search for variants that are in association with the efficiency of the therapy and the allergy, with the help of diffrent statistical programs.
I overviewed relevant literature, the basic statistical tools and methods, and the online genomic databases. Next, I summarized the allergen-SIT and the related mechanisms in the immune system in the human body.
I performed a prior-driven genetic association study by filtering the GWAS data to 37 genes, then I analyzed the corresponding 467 variants. Finally I examined the linkage disequilibrium between SNPs and their multivariate associations.
The main result of this prior-driven evalution was one SNP of the DEXI gene, with p-value<0.05 nominal significance, which was not genome-wide significant after the multiple hypothesis testing correction.
Using the results of the GAS, I have built artificial phenotypic models in the BayesEye program, then I generated synthetic data, which was used to investigate the power of the study and effect of sample size for plausible models.
With the help of PLINK program, I made quality filtering, then association tests concering the variants of 22 chromosomes. The result of these tests was two SNPs its Bonferroni p-value was genome-wide significant, but because of the preliminary status of the genetic data and the lack of correction for confounding variables in my analysis we can not yet conclude that they are associated with the ASIT response.
Furthermore I found variants that are in genes that are related with immunprocesses, but they do not have possess genome-wide significant p-values.
I have examined the results of the association tests with an online pathway analysis tool, Webgestalt, to support the gene-level and pathway-level interpretation of the variants.