genes reported in noUniprot can actually be found in data/uniprot.tsv

[PkiwiBird]

When running:
[ecModel, noUniprot] = makeEcModel(model, false);
there is this warning:

WARNING: The 30 gene(s) reported in noUniprot cannot be found in data/uniprot.tsv, these will
not be enzyme-constrained. If you intend to use different Uniprot data (e.g. from a
different proteome, make sure you first delete the existing data/uniprot.tsv file. 

1. Some of the missing IDs in noUniprot exist inuniprot.tsvfile, but are not detected as such maybe because there are several equivalent geneIDs split by ';', e.g.
   {'100137049'} exists in row P0C869 100137049;8681 3.1.1.4; 3.1.1.5 87978 MAVAEV....
{'100526794'} exists in row Q96P26 100526794;93034 3.1.3.5 68804 MSQTSL...
   How to solve this? Is it enough to split into two rows for each GeneID (2nd column) or would that create problems because the Entry (1st column) would be repeated?

2. Also, there is one gene {'6818' } that it shows in two rows:

P0DMM9	445329;6818	2.8.2.1	34196	MELIQ...
P0DMN0	445329;6818	2.8.2.1	34196	MELIQ...

Can this create problems?

3. There are Gene ids that are not in uniprot.tsv file. e.g. {'100288072'}
   How to deal with those?

Important parameters in ecModelGEMAdapter.m:

obj.params.complex.taxonomicID = 9606;
obj.params.kegg.ID = 'hsa';
obj.params.kegg.geneID = 'kegg';
obj.params.uniprot.type = 'proteome';
obj.params.uniprot.ID = 'UP000005640';
obj.params.uniprot.geneIDfield = 'xref_geneid';

[edkerk]

1. Duplicate Uniprot IDs are not allowed. You can maybe first check if both of the genes (100137049 and 8681) are included in the model. If e.g. 8681 is not present, then remove it from the uniprot.tsv. If both genes are present, then a work-around would be to change the gene association in the conventional GEM, e.g. replace all instances of 8681 with 100137049, and then again remove 8681 from the uniprot.tsv.
2. The first problem is of course that there are two genes (445329 and 6818) assigned to the same Uniprot ID. If you would remove 445329 from both of the entries, then GECKO would automatically choose the first protein it finds: P0DMM9.
3. In order to apply ec-constraints in relation to that protein, you should manually add the necessary information to the uniprot.tsv file. You might not raise a representative example though, as 100288072 seems to be in Uniprot and should be included with the UP000005640 proteome? Did you maybe have obj.params.uniprot.reviewed = true;? Try setting it to false.

[PkiwiBird]

3. Even after setting obj.params.uniprot.reviewed = false that 100288072 is in the noUniprot and doesn't exist in the uniprot.tsv

[edkerk]

No idea why it's not being picked up. Easiest solution might be to add the few missing proteins to uniprot.tsv yourself.