Comments made by J. Nogales, ph.D.
I really found GEMSiRV as a very complete and interesting tool for metabolic reconstructions.
Despite that by using models that you provide, the importing step was very straight forward; the greatest difficulty that I have found during the testing process has been to import no-standard models to GEMSiRV. I was unable to import successfully some models. Especially difficult was to import the recent model of Chlamydomonas published by our group and one of my unpublished models. After to use extensively “TextReplacer” I could import them, but some of the compartments were not recognizes by GEMSiRV. Specifically, the thylakoid was replaced by cytoplasm in all the attempts that I did. As consequence, I couldn't evaluate these models. I know that this is a critical step due the different source of the available models, but I’m wondering why I always found problems in models having thylacoid “[u]”. Response: This compartment thylakoid [u] has been added to GEMSiRV.
Besides these minor problems, I have found the rest of the applications included in GEMSiRV very interesting. The reconstruction module is very intuitive and friendly. The fact to can import rnx and metabolite databases significantly accelerates the reconstruction process. Also, I found very useful the mass and balance testing for each reaction. Although, similar applications such as rBioNet developed by Ines Thiele include this tool, only a few groups publish completely mass and balanced models. I understand that GEMSiRV is ready to publish, but since the reconstruction process is the main purpose of this tool, perhaps the inclusion of additional algorithms such as “gapFilling” and ”biomassPrecursorCheck” (both included in the cobra toolbox) could be very useful (the refinement process is clearly the limiting factor in metabolic reconstruction processes). On the other hand, I think that also could be interesting an individual output for the objective of the optimization process (e.g., BOF). Although for small model such as the E.coli Textbook this is not a problem, in a large genome-scale model is complicated looking for the result.
The visualization module is very interesting. I really don’t know how it can be improved. Without a doubt, the visualization is the main challenger for this kind of tools and GEMSiRV solves very well this particular task.
In summary, I think although additional algorithms could be included for the refinement process, GEMSiRV, together GBKPaser and MrBac will be an very useful tool in order to accelerate the metabolic reconstruction process.
Comments made by Y. H. Chen, ph.D.