Inconsistencies in chemical formulae

leuschjanphilipp commented 3 years ago

Hello,

with the new version of carveme(v1.5.1) I ran into inconsistencies with the chemical formula of some metabolites in my GEM. Does the new version solve mass imbalanced reactions as a MILP by adjusting the formulae (most visually in 'X' or 'R' containing formulae)? If so, is there any biochemical evidence on which this is build on?

A lot of metabolites in my model have a different formula as the one notated in BiGG:

Most of the time the difference is the amount of H-atoms
In some cases, the 'X' or 'R' term of the metabolite is substituted with e.g. C17H8O4
In some cases, the whole formula is just different: E.g. Chromate (BiGG id: cro4) has the formula CrO4 but in my model, it has the formula O2. This also happens in the grampos universe and also with different diamond parameters such as very- and ultra-sensitive. A model created with carveme 1.4.x does not have this problem and correctly assigned CrO4.

Could you please help me in understanding how the formulae (and their charges) are generated and if this solution has biological/chemical proof.

Kind regards, Jan

cdanielmachado commented 3 years ago

Hi Jan,

In previous versions CarveMe was using the formulae information retrieved from BiGG. However, there were often multiple formulae associated with a single metabolite. In the latest version there were two changes:

The formula (unique) is extracted from MetaNetX (instead of BiGG).
There is a new optimization method that tries to automatically balance reactions when the formulae are incorrect or inconsistent. It assumes that some formulae are incorrect and minimizes the number of formulae that it needs to change.

This is mostly an heuristic approach, so it is possible that it will also introduce mistakes in some formulae.

leuschjanphilipp commented 3 years ago

Hi Daniel,

thank you for your answer and explanation. Now I have a better understanding on where the formulae come from and how they are created during the carving process.

Also sorry for answering and closing this issue so late. Kind regards, Jan

cdanielmachado / carveme

Inconsistencies in chemical formulae #136