Templates stored within Polymer

[rwxayheee]

residue_chem_templates (ResidueChemTemplates) is an attribute of Polymer. The residue chem templates can be read from and written to JSON. Currently it includes a full set of available templates that has been loaded during the construction.

For each Monomer, there is also a template attribute but it is not a JSON-bound property (not exported to JSON or imported from JSON of a polymer).

Based on discussion we had today, we might want to re-evaluate whether we want and which templates to store with polymer. For templates generated during construction, this attribute serves as a useful cache. For templates that didn't match, retaining them may not be necessary.

[rwxayheee]

Closing as no change is needed and current behavior does provide useful information. The full set of templates makes it easy to re-run monomer matching.
For class Monomer, template and link_labels will stay as exceptions (#338) and not getting JSON serialized (?). Currently, they're only used during the building of padder_mol and can be re-computed as long as full set of templates is provided

[diogomart]

I think we should implement this, the serialized polymers take a lot of space and the meeko distribution contains the default templates anyway.

[rwxayheee]

@diogomart Can you implement this? I’ve thought about this, but I don’t know which attr will be used and how. I understand both are useful. If you can explain how you want this to be done (which one to keep, which one to remove) I can do the work

[rwxayheee]

Do you want polymer to have the set of templates it was prepared with? This might be different from the default, and might be useful to replicate the preparation for a new polymer or for the edited polymer

[diogomart]

That's a great point that the polymer might have been prepared with non-defaults. The monomer.rdkit_mol attributes are copies of the exact templates that matched (I think all info in the templates is preserved). Are you thinking of something more elaborate like preserving the templates that were associated in the ambiguous dictionary? For example, for a "HIE" residue, we'd keep also "HID", "HIP", "NHIE", ... , and so on.

Can you implement this?

Yes, I'll work on this, don't worry about it. It's low priority. But please still post your concerns about functionality we might be missing by implementing this.

[rwxayheee]

Hi @diogomart

Are you thinking of something more elaborate like preserving the templates that were associated in the ambiguous dictionary? For example, for a "HIE" residue, we'd keep also "HID", "HIP", "NHIE", ... , and so on.

Yes, I thought the full set would be useful if we want to re-parameterize this polymer, or to parameterize another polymer with the same exact set of templates, but I don't know if that's important and I agree that the object might be too big and not worth saving.

But please still post your concerns about functionality we might be missing by implementing this.

Not at the moment. I don't have a very clear visibility into where these attributes might be used outside of the library functions, so I’ll leave them as-is for now.