Merge Reactions With Same Stoichiometry But Different Genes

[Devlin-Moyer]

Main improvements in this PR:

Implements changes initially proposed in #580 and recently mentioned again in #630:

• Removes MAR02358 for being a duplicate of MAR08482
• Makes MAR08482 reversible Have compartment-specific GPRs for MAR08482 and MAR03931
• Removes MAR20066 for being a duplciate of MAR03984
• Removes MAR20020 for being a duplicate of MAR04521
• Removes MAR20002 for being a duplicate of MAR06631
• Removes MAR00030 for being a duplicate of MAR00025
• Merges annotations, references, E.C. codes, etc. from removed reactions into kept reactions
• Updates rxnRetired column for kept reactions in reactions.tsv

I hereby confirm that I have:

• Tested my code on my own computer for running the model
• Selected develop as a target branch
• Any removed reactions and metabolites have been moved to the corresponding deprecated identifier lists

[Devlin-Moyer]

I'm reasonably confident that the YAML validation failed because MAR00030 was the only reaction that ENSG00000072310, ENSG00000100979, and ENSG00000138823 were associated with; should I just go ahead and remove those genes as well? None of them appear to be enzymes, since MAR00030 wasn't really a "real" metabolic reaction.

[mihai-sysbio]

There's a pending PR that could be merged to develop which would add printing of the errors, thus increasing the certainty. Not sure how urgent this PR is - waiting a bit might help.

[Devlin-Moyer]

It's not urgent, and while that would be nice, using this

import cobra
model = cobra.io.load_yaml_model('model/Human-GEM.yml')
[m.id for m in model.metabolites if len(m.reactions) == 0]
[g.id for g in model.genes if len(g.reactions) == 0]

was pretty easy; the first list was empty and the second one was exactly the list of genes I mentioned above