GENE PRIORITIZATION BASED ON NETWORK AND FEATURE INFORMATION
MetadataShow full item record
High-throughput genome-wide studies such as sequencing or gene expression pro ling result in hundreds of potential candidate genes. Prioritizing these genes to fi nd potential candidate combinations, dominant molecular process, and causative sub-networks contributing to a disease phenotype is one of the most important problems of genomics. The progress in the understanding of molecular mechanisms under- lying common heritable disorders (e.g. autism, schizophrenia, diabetes) depends on the availability of new bioinformatics approaches for identifying characteristic genetic variations and associated multidimensional patterns of inheritance. In this thesis I address the problem of disease candidate genes and sub-networks prioritization and investigate the integration of broad signifi cant feature combinations, various models, speci fic networks, and novel algorithms, to achieve clean and highly con dent predictions. These can then be used to achieve a more effective and less side-effect prone drug design, and accelerate the process of understanding hidden disease models. To enhance speci ficity, both temporal and spatial context-dependent information is fused into the prioritization process. Besides gene-by-gene prioritization, I address the problem of prioritizing genomic sub-networks and gene sets, while considering the molecular interactions and clusters.