Abstract |
Every biological organism consists of cells. Cells are complex, biological systems whose
growth and existence depends on thousands of biological interactions between the molecules
(genes, proteins, chemical compounds etc). These interactions are typically modelled as graphs,
where nodes represent the molecules and edges the interactions between the molecules. Many
important biological networks have been discovered through laboratory studies. To date, at
least four types of biological networks have been characterized in detail: the gene regulatory
networks, the protein-protein interaction networks, the signal transduction networks and the
metabolic interaction networks.
This thesis focuses on the analysis and visualization of metabolic networks. Metabolic networks
represent the set of chemical reactions that happen in a cell to maintain life. Biochemical
reactions are processes that lead to the transformation of one set of chemical substances (called
reactants or substrates) to another set (called products). A variety of problems arise in analysing
and representing that knowledge using a graph-based model. For example, by treating each
metabolite in a reaction separately, dependences between metabolites are lost. Since a reaction
may have one or more substrates or one or more products, the edge representing a reaction is
actually a hyperedge. Thus, the metabolism is better represented by hypergraphs. A hypergraph
is a generalization of an ordinary graph where a hyperedge can connect more than two
vertices.
Unfortunately, the hypergraph based analysis and visualization of metabolic networks was
proven not a simple task, due to the complexity and the particularities of metabolic networks
and due to the lack of literature on hypergraph drawing. For testing our approach a tool, called
VisBolic (VISualization of MetaBOLIC pathways), has been developed. The tool analyses and
visualizes metabolic pathways while allows the execution of predefined but meaningful queries
on metabolism. The metabolic network data come from KEGG database, where the data are
stored in flat files. Therefore, the data were firstly preprocessed and then were stored in a
relational database, making them amenable to computer analysis.
|