Update BEL namespace to SCOMP and SFAM

[cthoyt]

I think for historic reasons, the SCOMP and SFAM namespaces were combine to BEL for the purposes of famplex. This is a bit problematic because the information about the original namespace is lost. Even more confusing, these are actually the names of the entities and not their selventa identifiers

I added all of the selventa namespaces to PyOBO in biopragmatics/pyobo#114. Would it break a lot of things in INDRA if we started updating these in famplex?

One compromise would be to add the additional references then just filter out the BEL namespace information in places that don't require it

[bgyori]

I think we're talking about different things. SFAM and SCOMP are actually relatively recent, and these mappings were done with respect to a version of BEL where they weren't used yet, hence the differences. "MAPK Erk1/3 Family" is a real BEL family that used to exist.

[cthoyt]

I saw the way this list was generated was by processing the selventa large corpus. I didn't realize that there's a older version of the large corpus that doesn't use SFAM and SCOMP!

[bgyori]

You can see in this file: https://raw.githubusercontent.com/sorgerlab/indra/de517e376256c61a3f91356bc8c95427febd8282/data/large_corpus.bel that the relevant namespaces that I used to import many families/complexes and curate mappings to FamPlex are PFH and NCH:

DEFINE NAMESPACE PFH AS URL "http://resource.belframework.org/belframework/1.0/namespace/selventa-named-human-protein-families.belns"
DEFINE NAMESPACE NCH AS URL "http://resource.belframework.org/belframework/1.0/namespace/selventa-named-human-complexes.belns"

At the time I didn't think these BEL namespaces were important beyond being able to map them when encountered. I then occasionally added things from SCOMP/SFAM later, again without splitting into a new namespace, to be able to process newer versions of Selventa BEL files where these appeared. I agree it would be good to update these since it's unlikely we will ever need to process old BEL content, we just need to be a bit careful that things stay operational downstream.