Modularize parameterization, joining topologies and creating systems

[ijpulidos]

Many protocols need to do the same things as parameterizing small molecule components, creating OpenMM topologies and merging them in a hybrid one, and creating OpenMM systems from that. Each of these should probably happen in their own callable and these should be ready to be used by the protocols as needed.

[IAlibay]

@ijpulidos we should discuss this more, this is already something we do in openfe and it's something I've been wanting to move out / do further work on.

[ijpulidos]

Yes, there are functions that do this in openfe, we are using some of them here in feflow. Initially I was thinking about refactoring this in openfe (in a separate branch or fork, as needed), because I know the migration of these could take more time.

[IAlibay]

🤔 I'm not sure I fully understand the intent. Ideally I would like us to not be in a situation where we are deciding long term code decisions this side of the year, is this something we can avoid?

i.e. is there anything right now that is necessary that is blocking other things?

I will focus my efforts towards refactoring the protocols in the new year, which will probably involve dealing with all this with all the stakeholders at once.

[ijpulidos]

The intention is not having to write code that will be redundant in the future. Especially if we cannot track exactly what would the redundant parts be. My thought was that maybe it is desirable that we can start working on this right now. But of course, I also don't want to put too much effort into this to later find out it wasn't the right approach, so I agree it needs further discussion, but I'm also willing to put some effort and find out if it fits in the future (next year).

[IAlibay]

That sounds good, thanks for agreeing to my selfish "can we wait until after the winter holidays" request!
That being said, maybe we can do some scoping next week so we can hit the ground running in the new year?

[ijpulidos]

Basically this issue and the PR that would eventually solve it would be my first proposal on how to solve this from my point of view. Which would be the accepted way or not in the longer term, that's fine, but hopefully it will be minimally functional that will enable having results on protein mutations, does that sound ok?

[IAlibay]

I think so, I will lean heavily on "I'm not sure I fully understand" 😅 - but erm as long as we're not blocking you that sounds great!