Dear researchers, I do not work directly on Flux Balance Analysis, but I studied the theory behind and I appreciate well done papers about the issue. I think that when made in the right way, this approach can provide useful hints. However, recently I found a number of papers where models are reconstructed for non-model organisms and it is never indicated the degree of overlapping with models available for model species.My feeling is that in most cases these new models are simply E. coli or yeast models, (slightly) adapted to the new species. This simply because there is not yet enough functional information for reactions not present in model organisms. Indeed TCA, pentose phosphate, glycolysis, amino acid biosynthesis and so on are more or less always present, but I wonder how much novel and species-specific reactions are included in such models beside the core Carbon metabolism that we can also model in yeast or E. coli. Given the absence of parameters in such models (except for minimum and maximum flux that are always set to a quite large range because we don't know the true range), if there are no specific reactions of a certain novel organism (or only periferic specific reactions, often with zero flux), one would obtain the same results with an E. coli model (eventually removing things not present in the novel organism). A similar issue arises concerning the objective function, which is more or less always similar also among very different organisms.
Anyone studied this overlapping and novelty issue more in detail?
If you haven't already seen it, you might find this article from 2013 somewhat helpful:
http://www.nature.com/nbt/journal/v32/n5/abs/nbt.2870.html
Hi Matteo.
I feel the answer to your first question is "always". The way draft models are generated (mainly using sequence similarity relationships) by automated tools usually generates coli-like reconstructions (at least in bacteria). So I would not be surprised to find that many (if not most) of the microbial models out there are just small modifications of the E. coli famous reconstruction.
The amount of novelty introduced in a new model is proportional to the effort put in manual curation after initial automatic reconstruction and to the available experimental data for that organism (see Thiele and Palsson bible on this: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3125167/). It is hard sometimes to find (or generate) such information for non model bugs but it is essential to produce something that brings some novelty to the community. Antibiotics production routes or fermentation processes are rarely accounted for by automated tools, for example. It is crucial to use papers and past information to refine the reconstruction and to render it bug-specific. By the way, this is often a great chance for bringing to new life papers that are dated back to the 70s, in which one can find accurate biochemical assays.
The very same concepts hold for biomass assembly reactions. A nice paper on this topic: http://link.springer.com/article/10.1007/s11306-015-0819-2
Unfortunately, the importance of reconstructions from non-model organisms is rarely recognised during the reviewing process (and no, I did not have a rejected paper on this issue recently!) but is arguably higher than flux predictions made on an ecoli-like model from a non e-coli like organism.
ciao,
m.
A quick update on this issue from a manuscript that I happened to find in hand.
Considering the following models, K. nataicola RZS01, E. coli K-12 MG1655, G. oxydans 621H, R. sphaeroides 2.4.1 (gamma and beta proteobacteria) a total of 137 genes (mostly belonging to the central carbon metabolism (8.2%), amino acid metabolism (24.1%), and biosynthesis of cytoskeleton (11.3%)) are common among the models.
not very coli-like models in this case.
m.
- Badges
- Science topic
- Similar topics
- Biological Science
- Molecular Biology