The symptoms of complex disease like allergy, obesity and cancer depend on the products of multiple interacting genes. High-throughput techniques have implicated hundreds of genes. There are also considerable individual variations. A clinical implication of this may be inadequate treatment response, which is increasingly recognized as a cause of increased suffering and costs. Ideally, physicians should be able to routinely personalize medication based on a few diagnostic markers. Finding such markers is a formidable challenge. We hypothesize that translational clinical studies based on high-throughput genomics, advanced computing and systems biology may help to identify markers for personalized medication in complex diseases. We organize disease-associated genes in networks that are analyzed in a top-down manner. First, modules of interacting genes with distinct biological functions are identified. Then the modules are dissected to find pathways and finally upstream genes with key regulatory functions. We use bioinformatic methods that were recently described by us in Nature Genetics and Nature Biotechnology. An important focus of this project is to develop these methods to form multi-layer modules that integrate information about disease-associated changes on the DNA, RNA and protein levels. Since these levels interact, studies of the different levels can be interactively used to cross-validate the modules. This involves both genetic and experimental studies, but the ultimate test of the modules will be if they can be used for clinical predictions. For example, changes in RNA expression may be caused by a single nucleotide polymorphism (SNP) in a regulatory region. If so, the corresponding protein is tried as a marker to personalize medication. We have chosen hay fever as a model of complex disease because it is common, well-defined and readily examined in clinical and experimental studies. However, the methods may be generally applicable to complex diseases.