Be careful about subject/object distinctions in Monarch mapping commons?

[matentzn]

@kevinschaper mentions in his update:

I'm not quite sure how to think about the SSSOM being built of triples, and that those factual statements don't really dictate what ID we want or don't want, and that direction shouldn't really matter. In my code, I really want one column that is the ID I want and the other column to be IDs that I might potentially have, and I want to apply them with simple joins.

I think this is a great prompt for thinking about how to realise this. A few days ago I requested a "flip" operation in sssom py (flipping is not trivial due to metadata). I believe we can achieve perfectly formatted mapping sets for monarch with a mix of flip and filter.