As anyone who has tried to reproduce a published mathematical model will testify, it’s a long, tedious, and generally futile task. Equations are replaced by ambiguous descriptions, parameter values are left undefined and, worst of all, the main author has given up science to set up a vegan cup-cake business.
And so Natalie Stanford and I set Holly Summersgill and Ed Kent (two talented students from the Systems Biology DTC) this long, tedious and futile task, with four models of the pentose phosphate pathway:
- Haut, 1974: Simulation of the Pentose Cycle in Lactating Rat Mammary Gland. PMCID:PMC1166237
- Sabate, 1995: A model of the pentose phosphate pathway in rat liver cells. PMID:7753046
- Vaseghi, 1999: In vivo dynamics of the pentose phosphate pathway in Saccharomyces cerevisiae. DOI:10.1006/mben.1998.0110
- Ralser, 2007: Dynamic rerouting of the carbohydrate flux is key to counteracting oxidative stress. DOI:10.1186/jbiol61
Results were a mixed bag. Two of the models were fully reproducible, a third was close enough. In the Sabate model, however, the equations and results seemed totally inconsistent, differing in all cases by orders of magnitude. Answers on a postcard please.
But the project was about more than just reproducing results; we also aimed to disseminate the models to allow their easily re-simulation by the community. For this two acronyms were required:
- SBML: a standardised format for representing models of biological processes, supported by many software packages (including Matlab, Mathematica, …)
- MIRIAM: a standardised format for annotating the entities of those models. By marking-up the molecule “gluc” as CHEBI:17925 allows its unambiguous identification and automatically links to many additional sources of information.