Templates

[JamesJeffryes]

This PR updates the templates by replacing reactions marked as obsolete in the template with a valid linked reaction. Also, reactions that were not atom balanced were removed from the template. (only 9 from core & gram +&- but many more from human & mycobacterial templates) The Validate Template file has been updated to check for obsolete and imbalanced reactions on travis.

I've requested review from those that might have an opinion on changes to the templates.

Fixes #48

[mmundy42]

For reactions that have a type of “gapfilling” it should be fine to remove them from the template. But I did notice that some of the reactions have a type of
“conditional” which means they are linked to a complex and a role. I would prefer to keep conditional reactions since they improve the reconstruction
of draft models.

[mmundy42]

Yes, that is a good compromise. We can review later and figure out if the troublesome reactions can be replaced by reactions that have OK status.

Does this sound like a good compromise? I would change the validation to allow imbalanced reactions if they are conditional.

[samseaver]

Unbalanced reactions should never be allowed in the template, but if there are conditional reactions that are unbalanced, they should either be fixed directly, or replaced with the appropriate balanced reaction.

Since these files have not yet been updated from the current objects being used in production, I have no problem removing them for the time being so that the checks pass, but we should maintain a separate file (added to this PR if possible) that lists: The reaction being removed and the template it was found in.

When time comes to actually maintain the templates through this repository, we'll have to curate these.

[janakagithub]

James, do you save those unbalanced reactions from core/gram-/+ templates? I think some of those are key reactions (may be all the reactions), those need to be atom balanced and added back to the template.

[JamesJeffryes]

Here are the reactions that are imbalanced. Some reactions like DNA synthesis cannot be balanced but we can change the status to OK if we want to keep them.
imbalanced rxns.txt

[JamesJeffryes]

@janakagithub @jplfaria Do you want to attempt to balance these reactions?

[chian]

Just as a note, I don't think many of these reactions are not going to be balanced because they are sinks into concepts like "protein synthesis ". And they involve fractions of things. Also, even if you do give a concept a molecular formula, it would eventually ruin the point as something like protein synthesis is probably meant to change amino acid ratios.

…

Sent from my iPhone

On Aug 28, 2017, at 3:52 PM, James Jeffryes ***@***.***> wrote: @janakagithub @jplfaria Do you want to attempt to balance these reactions? — You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub, or mute the thread.

[JamesJeffryes]

So I changed approach on this. The Mayo Clinic templates are exempt from validation while I was able to update or balance reactions for all but rxn01659. This reaction produces a plant hormone and is only used for gapfilling so I think it's reasonable to drop.

[samseaver]

The change to formula in the Fatty-acyl-ACPs isn't right. I'd know because I tried to systematically change them all some time ago. We need to explicitly include the formula for the 4-Phosphopantetheine side chain because it's a key cofactor biosynthesized directly from a vitamin.

You'll find that all fatty-acyl acps, as far as I could, have this formula, and as such, if you have unbalanced reactions, then you need to change the formula of the fatty-acyl acp that is on the other side of the reaction (which is in fact common to both reactions here, so, evidently, I missed one)

[JamesJeffryes]

So I'm going to close this PR and make the changes in the JSON files