Ralf Hofestädt
Ralf Hofestädt
Bielefeld University, AG Bioinformatics/Medical Informatics
hofestae@techfak.uni-bielefeld.de
More than 500 database systems are available which represent molecular data. Therefore, experimental data and experimental results of fundamental metabolic processes like gene regulation, metabolic pathway control, signal pathway control and cell differentiation processes are available via the internet (Collado-Vides et al 2002). The actual condition of these systems is that most of them can be characterized by an exponential growth. However, for the analysis of complex metabolic networks only rudimentary data and knowledge is available today. Therefore, we have to develop and implement special algorithms for the analysis and synthesis of complex metabolic networks, which are able to complete the rudimentary data. In practice a scientist has rudimentary knowledge about the metabolic scenario he is working with. The first step to accumulate this knowledge will be a complex database query. This can be done manual or using database integration tools like SRS.
The second step is to use algorithms for this accumulating process, which allow the prediction of metabolic networks. To realize the data query process different integration tools are available behind the SRS system, which support the integrated data query process. For the prediction of metabolic pathways powerful tools are still missed. However, these previous approaches and existing metabolic pathway databases have a number of limitations in metabolic pathway reconstruction.
Knowledge of the genome sequence alone is insufficient. Predicting each gene function based solely on sequence similarity searches often fails to reconstruct cellular functions with all necessary components. They do not contain comprehensive information about metabolic pathways, such as physical and chemical properties of the enzymes that involved. Some are not fully computer-aided. The individual database search process requires too much human intervention and the quality of annotation largely depends on the knowledge and work behavior of human experts. The future of metabolic pathway analysis may depend upon its ability to capitalize on the wealth of genetic and biochemical information currently being generated from genomic and proteomic technologies. An ideal system for metabolic pathway prediction would include a Web-based architecture to allow remote and local access to the different biological databases. It would offer a proven approach that can perform complex queries, data transformations, and data integration under one simple interface, without requiring extensive programming.
