Since March 2008, I work as a postdoctoral researcher at the department of Molecular Genetics at the University of Antwerp, in Antwerp, Belgium. This department is part of the Flanders Institute for Biotechnology.
Together with linguists and data miners, our current research project aims to develop data mining techniques that abstract from scientific literature in order to find, e.g., suitable genes with a high plausibility of having an effect on disease.
In order to accomplish this goal, academic content (as in relations between biological elements, processes, structures, ...) has to be text mined from the literature relevant to a specific disease or gene. Secondly, all of this data needs to be collected and organized in a workable format (graph, DB, ...). Thirdly, data mining algorithms have to find non-obvious but significant relations in the resulting graph, which can then be tested in the lab.
Over the last few years I've put some research into building a web service for the discovery of biomedical knowledge in existing, published knowledge. A common problem for many researchers in biomedicine is that they get lost in the published literature, but more recently also get lost in the huge amount of data and putative research targets that are generated by their high-throughput experiments. Biograph combines a lot of knowledge bases from the biomedical domain into one central knowledge network and adopts mining techniques to discover relevant and specific links among likely targets. As a result, Biograph can help determine which of these targets is most interesting and automatically generate functional hypotheses to ground these rankings. Biograph is available via http://www.biograph.be