Incorporate Protein Memberships

[cthoyt]

I'm not really sure where to begin on this. First step would be to check the AETIONOMY code on gitlab to see what @cebel was able to load before.

Possibilities:

• InterPro XML contains much more rich information
• UniProt has cross-references to InterPro so it would be possible to use PyUniProt to get some of this information

[aramgrigoryan]

can you give a bigger picture on this? like what\s the basic idea behind this? what is the original problem you wanted to solve? and what's the final thing that you wanted to get?

[cthoyt]

Which proteins belong to which families? The interpro hierarchy doesn't have this, but interpro does somewhere else. There are more files in the interpro dump, and we might have to spend some more effort to parse them up and get it in our database models

[cthoyt]

Actually it will require making a new database model for proteins and also for connections to proteins.

Then, we will serialize these relations to bel. Then we will write an enrichment function for bel graphs

[aramgrigoryan]

didnt i do that with this? ftp://ftp.ebi.ac.uk/pub/databases/interpro/ParentChildTreeFile.txt

[cthoyt]

When we parsed this and added it to a NetworkX graph, it's not using the data structure of the bel graph. Also, this function should only add relevant parts of the interpro hierarchy to Ann already existing Bel graph. Check the unit tests

[aramgrigoryan]

Does this tell something? so we can look for type=family