search

SBRG / bigg_models

The BiGG Models website server
http://bigg.ucsd.edu
Other
77 stars 18 forks source link

Discrepancy in E. Coli core #324

Closed tjclement closed 5 years ago

tjclement commented 5 years ago

Description of the issue

In E. Coli core, the "Formate transport via diffusion" reaction is marked as reversible in the SBML file, whereas it is marked as irreversible in the Matlab file. I assume one of the two is erroneous.

Page

http://bigg.ucsd.edu/models/e_coli_core

Browser

Mozilla/5.0 (Macintosh; Intel Mac OS X 10_14) AppleWebKit/605.1.15 (KHTML, like Gecko) Version/12.0 Safari/605.1.15

zakandrewking commented 5 years ago

Thanks for reporting. I see the issue: When a reaction is irreversible but in the reverse direction, then the SBML reversible attribute is true instead of false.

@draeger Looks like we could fix this in ModelPolisher?

draeger commented 5 years ago

This is a bit tricky. The concept of reversibility works differently in COBRA modeling and in SBML. While a COBRA model can have an irreversible reaction that proceeds effectively in reverse direction, this is not directly possible in SBML. There, irreversible reactions can only proceed in the forward direction. Two solutions are possible:

  1. Flip reactants and products in the SBML export and flag the reaction as irreversible.
  2. Set reversible = "true" in SBML but have an upper bound of 0 and some non-zero lower bound to effectively express that the reaction only proceeds in reverse direction.

If we go with solution 2, we cannot set reversible="false" because this would conflict with the definition of the upper and lower bound. Now, @tjclement, could you please check if the SBML version of the model under dispute uses already solution 2?

tjclement commented 5 years ago

Thank you for the quick responses.

@draeger Indeed, the model has a lower bound of -1000, and an upper bound of 0.

draeger commented 5 years ago

Thanks for checking. As I see it, that's the only "trick" we can do in the SBML version, unless flipping reactants and products is acceptable, too. I think we can mark this issue as solved because this is just about how SBML is defined. But thanks for pointing this out! Maybe this solution for the SBML export should be communicated in some central FAQ somewhere.

tjclement commented 5 years ago

I understand. I'll close the issue, thanks.