| Molecular biologists are gradually assembling the structures and constituents of numerous biological networks. These networks represent interactions that take place within a living organism between proteins and other biologically significant molecules. This thesis presents a new algorithm for producing a global alignment of two given metabolic or signalling networks. Alignments of networks expose conserved subnetworks, which in turn can reveal how the networks may have evolved. Also, network alignments can be used to measure the similarity of the biological systems they represent.;The alignment is produced through a matching of the most similar nodes from the two networks, with the similarity based on comparison of the biological molecules at the nodes being compared, as well as the local network around those nodes, referred to as their neighbourhoods. The project evaluates alignments of networks from the Kyoto Encyclopedia of Genes and Genomes (KEGG) produced by different variations of the algorithm and compares them to baseline alignments based on the similarity of the molecules at network nodes alone, without taking any local network structure similarity into account. Different alignments are evaluated based on how well they agree with alignments implied by KEGG reference networks. The results show that the complete algorithm performs much better than matching based on node sequence similarity alone for aligning both metabolic and signalling networks. |
