PhD Lecture by Kasper Lage Hansen, CBS

Bioinformatics and systems biology of human diseases

Thursday 14 August 2008 at 14:00
CBS, DTU, Lyngby, Building 306, Auditorium 31
Assessment Committee:    Associate professor Ramneek Gupta, DTU (Chairman)
Head of Square One Søren Schandorff Gade-Jørgensen, University of Southern Denmark
Professor, dr. Gert-Jan B. van Ommen, Leiden University Medical Center
Chair of defense:Professor Ole Lund, DTU
Supervisor:Professor Søren Brunak, DTU

Thousands of protein interaction experiments across many different organisms have provide us with the opportunity of studying large systems of interacting proteins. These analyses have shown that protein interactions follow universal rules that also apply to other networks for example the internet and social networks. This inherent property of proteins can be exploited by computational methods to systematically deduce functional relationships from these massive amounts of data. Here, our work uses protein interaction data as the backbone in a number of integrative analyses focused on human cellular systems in health and disease. By combining interaction data with dynamic- and sub-cellular localization data we analyze the functionality of a complete cellular compartment the nucleolus. This provides new insight to a number of biological processes for example the ribosome biogenesis pathway in humans. Then, in two different analyses we use interaction data in combination with other data types to identify cellular systems highly incriminated in the aetiology of type 1 diabetes. These two distinct analyses point to novel specific candidate proteins in the disease. We also provide a systems view of the processes taking place in insulin producing pancreatic beta-cells, after these cells are attacked by infiltrating immune cells. These attacks eventually lead to apoptosis of the beta cells, lack of insulin secretion and type 1 diabetes. Our work further shows how text-mining of human disease descriptions can be used to generate a human disease network, a phenome. By integrating the phenome with data on the interactions of proteins known to be involved in diseases we make the first large-scale analysis of protein complexes implicated in human genetic disorders. This analysis also provides a list of 114 candidate genes in a wide range of diseases. Finally, we add spatial resolution to the analyses of disease genes and disease protein complexes, by using an array atlas of tissue-specific gene expression. Using text-mining we further systematically map cancer and non-cancer diseases to the tissues most affected by the particular pathologies. By intersecting these data-dimensions, we are able to make an organism-level analysis that provides new surprising insight to the biological processes that prime certain tissues for specific tumor formation.

Everybody is welcome. Registration is not necessary.