Consider removing generic protein (de)phosphorylation reactions

[JonathanRob]

Note: This Discussion topic was resolved in #337.

Human-GEM currently contains a few generic protein phosphorylation/dephosphorylation reactions that are each associated with many genes:

Reaction ID	Equation	# Genes
MAR09577	ATP [c] + [protein] [c] ⇒ ADP [c] + phosphoprotein [c]	351
MAR09578	H2O [c] + phosphoprotein [c] ⇒ H+ [c] + Pi [c] + [protein] [c]	78
MAR09579	ATP [c] + protamine [c] ⇒ ADP [c] + O-phosphoprotamine [c]	212
MAR07617	ATP [c] + [protein]-L-tyrosine [c] ⇒ ADP [c] + H+ [c] + [protein]-tyrosine-phosphate [c]	113
MAR07618	H2O [c] + [protein]-tyrosine-phosphate [c] ⇒ Pi [c] + [protein]-L-tyrosine [c]	86

These reactions are all dead-ends, which is not a problem in itself, but they are associated with a very large number of primarily kinase-encoding genes. I am generally in support of trying to add more information to Human-GEM when possible, but it is important to keep in mind that as a metabolic model, it is often necessary to limit the scope of the GEM to metabolic processes. The definition of what is "metabolic" is relatively imprecise and will vary depending on who you ask, though I would argue that these hundreds of kinase-encoding genes are not so "metabolic" and fall more into the "signaling" category. They also do not contribute much information to the model, as they are all attached in a giant group to a single type of chemical transformation (addition/removal of a phosphate group).

I suggest that we remove these 5 reactions from Human-GEM, to avoid having so many generic kinases that do not participate anywhere else in the metabolic network. @johan-gson has also recommended this solution in the past.

What do others think?

[johan-gson]

I'm am in favor of this. It is clear to me that the GPRs are meaningless. What I don't fully understand is if the reactions are needed for something. But if they are dead-end, maybe they are not.

[haowang-bioinfo]

agree, inclusion of over generic reactions involving a lot of genes is not much useful to a GEM. Sometimes, this may give misleading results.

[mihai-sysbio]

How about mentioning this in the Gitter channel and then creating an issue for this?