Strange differences between text files in bin and XML file

[akaviaLab]

There is at least one reaction that has strange differences between the text files in bin and the XML file MODEL1603150001.xml. Specifically, the reaction R_FAOXC220200x (there might be others, I'm searching for them).

If we look at this reaction in the TXT files, it appears three times (once for genes, irrelevant for now, twice with different formulas)
mcisb-recon-master/bin/reactionsToAdd.txt:R_FAOXC220200x R_FAOXC220200x M_h2o_x + M_coa_x + M_o2_x + M_nad_x + M_c220coa_x --> M_h_x + M_h2o2_x + M_nadh_x + M_accoa_x + M_arachcoa_x
mcisb-recon-master/bin/reactionsToUpdate.txt:R_FAOXC220200x M_h2o_x + M_coa_x + M_o2_x + M_nad_x + M_c220coa_x + 3.0 M_h_x --> M_h2o2_x + M_nadh_x + M_accoa_x + M_arachcoa_x

However, in the XML file, it has a different formula, where it has 3 H+ as reactants and 4 H+ as products (relevant XML pasted below).
M_coa_x + M_o2_x + M_nad_x + M_docoscoa_x + 3* M_h_x --> 4*M_h_x + M_h2o2_x + M_nadh_x + M_accoa_x + M_arachcoa_x.

What is going on here?

---

  <reaction id="R_FAOXC220200x" name="R_FAOXC220200x" metaid="_4a255025-baf1-4c0f-ade9-b1fe265564f7" reversible="false" sboTerm="SBO:0000176">
    <notes>
           <html:body xmlns:html="http://www.w3.org/1999/xhtml">
              <html:p>GENE_ASSOCIATION: (HGNC:10606 and HGNC:119 and (HGNC:3247 or HGNC:5213)) or (HGNC:119 and (HGNC:3247 or HGNC:5213) and HGNC:82)</html:p>
           </html:body>
        </notes>
    <listOfReactants>
      <speciesReference species="M_h2o_x" metaid="eb81c5a4-33eb-4f5e-ab96-f7ca0ac570db" stoichiometry="1"/>
      <speciesReference species="M_coa_x" metaid="ae245fe7-e7c4-4602-a15b-e6d47b1c31f6" stoichiometry="1"/>
      <speciesReference species="M_o2_x" metaid="e0525429-aea2-4996-8f6d-87dd6ad2e6fc" stoichiometry="1"/>
      <speciesReference species="M_nad_x" metaid="_080440c6-16f0-4259-bae6-f59bbacef0e0" stoichiometry="1"/>
      <speciesReference species="M_docoscoa_x" metaid="_77e8ff2b-88e5-476f-baa1-017c364018a8" stoichiometry="1"/>
      <speciesReference species="M_h_x" metaid="ccc7525b-455c-41e8-821c-42854e04752b" stoichiometry="3"/>
    </listOfReactants>
    <listOfProducts>
      <speciesReference species="M_h2o2_x" metaid="c84fedc4-c3d9-41fe-a083-519ed7b85303" stoichiometry="1"/>
      <speciesReference species="M_nadh_x" metaid="b847cee7-a762-4b5d-9c04-1d40f3bc9b24" stoichiometry="1"/>
      <speciesReference species="M_accoa_x" metaid="cde1c5f7-799c-49e0-b0b4-5755d0d06cf0" stoichiometry="1"/>
      <speciesReference species="M_arachcoa_x" metaid="_5398518d-ffa0-42c6-8224-495d1e2db68f" stoichiometry="1"/>
      <speciesReference species="M_h_x" metaid="_34be2c1e-3d4d-4e7a-bffd-464a29176668" stoichiometry="4"/>
    </listOfProducts>

[neilswainston]

Good spot! There is indeed some redundancy here.

The reactionsToAdd file is read first, and the reaction (obviously) added to the model. Later, reactionsToUpdate is read, and the reactions updated. In an ideal world, the reactionsToAdd would have specified this reaction correctly, meaning that the update is not necessary. This is (probably) an oversight based on the confusion of curating a model with thousands of reactions.

The final model contains 3 h_x reactants and 4 h_x products due to the auto-balancing method, which attempts to correct for missing water, protons and incorrect stoichiometry by applying linear programming to the task of balancing (see http://biecoll.ub.uni-bielefeld.de/volltexte/2011/5202/pdf/jib_186.pdf). The final reaction is therefore correctly mass / charge balanced.

However, I appreciate the confusion here, as the initial reaction is (as far as I can see) mass and charge balanced:

R_FAOXC220200x R_FAOXC220200x M_h2o_x + M_coa_x + M_o2_x + M_nad_x + M_c220coa_x --> M_h_x + M_h2o2_x + M_nadh_x + M_accoa_x + M_arachcoa_x

Essentially, the reactionsToUpdate update (and subsequent auto-balancing) is probably superfluous.

I'm sure that there will be other anomalies like this. However, the key take-home is that all reactions (except biomass production) are mass and charge balanced - hence the sensible ATP flux predictions.

[akaviaLab]

Well, the original metaoblite c220coa is replaced with M_docoscoa_x, which is supposed to be identical. Maybe the script got confused at some point during the replacement process? It is also a bit weird, since I think the reaction FAOXC22C20x is already present in RECON 2.04, see https://vmh.uni.lu/#reaction/FAOXC22C20x (and reaction index 7054 if you're looking at the model file). I'm just trying to figure out how common this kind of occurrence is, and if it needs to be fixed. Cheers, Uri David

…

On Wed, Mar 8, 2017 at 4:02 AM, Neil Swainston ***@***.***> wrote: Good spot! There is indeed some redundancy here. The reactionsToAdd file is read first, and the reaction (obviously) added to the model. Later, reactionsToUpdate is read, and the reactions updated. In an ideal world, the reactionsToAdd would have specified this reaction correctly, meaning that the update is not necessary. This is (probably) an oversight based on the confusion of curating a model with thousands of reactions. The final model contains 3 h_x reactants and 4 h_x products due to the auto-balancing method, which attempts to correct for missing water, protons and incorrect stoichiometry by applying linear programming to the task of balancing (see http://biecoll.ub.uni-bielefeld.de/volltexte/2011/ 5202/pdf/jib_186.pdf). The final reaction is therefore correctly mass / charge balanced. However, I appreciate the confusion here, as the initial reaction is (as far as I can see) mass and charge balanced: R_FAOXC220200x R_FAOXC220200x M_h2o_x + M_coa_x + M_o2_x + M_nad_x + M_c220coa_x --> M_h_x + M_h2o2_x + M_nadh_x + M_accoa_x + M_arachcoa_x Essentially, the reactionsToUpdate update (and subsequent auto-balancing) is probably superfluous. I'm sure that there will be other anomalies like this. However, the key take-home is that *all* reactions (except biomass production) are mass and charge balanced - hence the sensible ATP flux predictions. — You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub <#3 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AKGMZoMfftDlV63MwdGvQdsr_j3Opoydks5rjm6cgaJpZM4MUrp_> .

-- Uri David Akavia, Ph.D. Assistant Professor, Biochemistry Department [email protected] McIntyre Medical Sciences Building 3655 Promenade Sir William Osler, Room 805B Montreal, Quebec H3G 1Y6 Office tel: 514-398-8526 fax: 514-398-7384