No EC number in ec.eccodes although EC is in uniprot.tsv

[PkiwiBird]

There's one reaction in a model that has an entry in ec.enzymes: 'A6NKP2' and it is present in uniprot.tsv file:
A6NKP2 100288072 1.1.1.- 46869 MKSNPPRSSLEACKAAGQAPQQKTQAK
but it is empty in ec.eccodes after running:
model = getECfromDatabase(model)
Then it gets a kcat value of 0 after running:

 kcatList_fuzzy  = fuzzyKcatMatching(model)
model  = selectKcatValue(model, kcatList_fuzzy)

There's also NaN in .origin of kcatList.
Could it be that the corresponding entry in ec.eccodes is empty because the EC, 1.1.1.-, is not complete? and therefore does not find a kcat?
In that case, does it mean the fuzzy kcat matching only works after getting a complete EC?
If so, would be interesting to get gecko to get the fuzzy kcat matching (for a similar enzyme, or same enzyme other substrate, etc) with an incomplete EC?

[edkerk]

The behaviour of fuzzyKcatMatching and selectKcatValue are as expected and intended, but getECfromDatabase should also fill incomplete EC codes such as 1.1.1.-. In full_ecModel/models/ecYeastGEM.yml you can for instance find this example:

  - !!omap
    - id: "r_2773"
    - kcat: 13.8
    - source: "brenda"
    - eccodes: "2.3.1.-"
    - enzymes: !!omap
        - Q06510: 1

So in your case it seems that there is something going wrong with finding the entry in uniprot.tsv, which is done by comparing the ec.genes field with the gene identifiers (2nd column in uniprot.tsv). That this fails is somewhat puzzling to me, as makeEcModel apparently did not have a problem to find the same gene, otherwise A6NKP2 would never have ended up in ec.enzymes. Do you by any change use a data/uniprotConversion.tsv file? There might be a bit of difference how getECfromDatabase and makeEcModel deal with those, so maybe that is the issue.

Also, do you a kegg.tsv file? Because if the different handling of data/uniprotConversion.tsv is indeed the problem, then this likely means that the EC values that are ending up in your model are all derived from KEGG and not UniProt (in other words, the problem is not unique for this particular enzyme, but none of the enzymes get their EC from UniProt due to a bug in the mapping).