I would like to superimpose metabolic flux balance analysis results on KEGG pathway maps and would like to know if there is a way to systematically associate BIGG metabolic model reaction IDs to KEGG enzyme/reaction IDs. Some reactions have EC numbers (e.g. those for E. coli iAF1260), but it appears that those reactions account for a relatively small percentage of the total number of reactions and, in some cases, multiple reactions can have the same EC number, so it's not entirely clear how best to associate BIGG model reactions with those particular ones in the KEGG pathway maps. Any suggestions on how to systematically implement (and potentially automate) this mapping will be appreciated.
This is a complicated problem because the level of annotation of the models in the first place is not very good. What we do is that we start by mapping the small molecules present in BiGG models to referenced entries in a database (like ChEBI, PubChem or KEGG) either by provided references, which tend to be scarce, or by name. After this, what we do is map those referenced molecules to reactions, which doesn't mean that you always find an exact reaction in KEGG that matches. At the reaction comparison level you need to be flexible as well (ignoring differences in Hydrogen atoms and water, as some databases sometime add these to balance protons). As you well said, EC numbers sometimes point to slightly different reactions, so the best is to try to compare reactions as we do on the small molecule level. You'll find that most of the work is actually on correctly mapping the small molecules present in BiGG models to entries in databases, and at least in our experience this requires a level of manual intervention.
Now, if you would like something out of the box, I could mention (not sure if recommend, as we haven't evaluated the quality of the resource yet) the MetRxn database, where they have tried to unify BiGG models with other metabolism databases:
It currently says it is down for maintainance, but last time I tried it it gave results that you might find useful. Wait some days and give it a try.
There is paper I think for MetRxn.