Sammendrag
The fish farming industry is expanding, and to achieve economic and ecological sustainability, new fish feeds are being developed. When developing new feeds, it can be useful to first simulate on a computer how the biological network will react. This can be done with metabolic models. Metabolic models consist of the reactions and metabolites arranged into a stoichiometric matrix. Constraints on the network are imposed in the form of stoichiometric coefficients and bounds on reaction rates. To trust the results from a simulation, it is important that the model is well annotated and internally consistent, i.e. of high quality. The software Memote tests the model for a set of quality criteria and presents the score in a report. This thesis will discuss the application, development and validation of Memote’s tests. This is done by implementing three possible improvements iteratively to a model and testing with Memote for each iteration, eventually composing a Memote history report to inspect the change in score for the different model versions. There is an overall emphasis on annotations to databases in the tests; a wide range of annotations for genes, metabolites and reactions will increase the score. Including physiologically important reactions, such as secretion of CO2 will also increase the score. Memote is a good tool to show what the model contains, the scope and notify you if a feature you think was added to the model, was in fact not added. To more thoroughly review the models, adding organism-specific tests could be a possibility.
Vis fullstendig beskrivelse
