Biometrical support for the priority program "Immune and Metabolic Modulation induced by severe Tissue Trauma" and data mining approaches for the identification of disease associated single nucleotide polymorphismus (SNPs)

The first goal of this project is the biometrical support of genetic epidemiological projects within the priority program ¿Immune and Metabolic Modulation induced by severe Tissue Trauma¿ with a focus on the project ¿Genetic Predisposition for Adverse Outcomes in Trauma Patients¿. Here, the co-operation will comprise all biometrical aspects starting from quality control via statistical analyses up to the interpretation of results with regard to genomic controls, the group sequential study design, single marker and haplotype analysis, gene-gene and gene-environment interactions. The aim is to yield a validated model to predict the occurrence of adverse outcomes in trauma patients from a parsimonious number of genetic and environmental factors. The second goal is the methodological comparison and further development of different data mining algorithms for identifying genetic susceptibility SNPs and gene-gene interactions (epistasis). Emphasis will be put on two nonparametric approaches that have been developed recently, namely, the multifactor dimensionality reduction (MDR) method and random forests (RF). These will be compared with regard to accuracy, interpretability, and ease of application.