Refactor: mondo.sssom.config.yml.subject_prefixes -> metadata/SOURCE.yml

[joeflack4]

Overview

The subject_prefixes section of mondo.sssom.config.yml exists, as far as I'm aware, in order to tell us (particularly to tell lexmatch), what prefixes pertain to term classes. This allows lexmatch to filter and match only on the correct things. I'm not 100% sure, but I think it might be best to move this information into the metadata/SOURCE.yml.

Questions

1. Redundancy?

This section feels redundant, or perhaps misplaced. There are several places now where prefixes need to be added when we add a new source to Mondo. Some of these are automated by curies, e.g. the OBO context JSON and now the SemSQL prefixes.csv. But I think this needs to be handled separately from curies. Perhaps this isn't really a question. As I'm thinking about it, I don't see where this is redundant. This list subject_prefixes represents the subsets of prefixes in an ontology (classes) that we want to do matching on.

2. Best location?

I was thinking that maybe the best location for this would be in each ontology's metadata/*.yml file. We have a prefix_map and a base_prefix_map (subset of prefix_map that is 'owned' by the ontology) there. It would seem to me that a good place to put this. Isn't it the case that the prefixes we have in subject_prefixes in mondo.sssom.config.yml represent things we want to match against, which would only be classes, right? If that's the case, then should we not have a class_prefixes section in metadata/*.yml and put this there instead? Intuitively also that just feels like the best location to me. Though if this is only going to be used by mondo.sssom.config.yml, maybe it is indeed best there. Not saying this is a big priority; if it's not broke, don't fix it, as the saying goes. Thoughts?

[joeflack4]

@matentzn I created this based on #434 (comment). I agree this is best as a new issue.
@twhetzel We discussed this in a call. No need to worry about this this is super low priority.
@hrshdhgd You're the most relevant person to discuss this with me, but this is super low priority.

[hrshdhgd]

So you're basically suggesting just to rename the file? or extract out the subject_prefixes part to a new file?

[joeflack4]

Suggesting this one:

or extract out the subject_prefixes part to a new file

Pros:

• Everything related to understanding source's prefixes and what purposes they serve remain in their respective metadata/SOURCE.yml

Cons:

• Requires refactoring of lexmatch, and also introduces new complexity (adding metadata/SOURCE.yml as another input)

[hrshdhgd]

If it's as minor as that, got ahead and do it. As long as it doesn't break anything 😄 . I don't have strong opinions

[matentzn]

Ithink subject_prefixes may be historical, and only relevant in the mondo repo. Feel free to remove it and see if the pipeline still runs as @hrshdhgd suggests!

[joeflack4]

It wouldn't be a removal; it'd be a refactor. It's a low priority but I'm glad we came to a determination here! Good to know that this is also something that is not expected as some sort of standard SSSOM config.