Copy new intercell related procedures from the R package

[deeenes]

I am just writing this as a note, and once @michalk8, me or someone else will have a little time, it would be great to implement these methods here.

Based on the experiences of working with cell-cell communication, we created these new solutions in OmnipathR:

• Quality filters for the OmniPath intercell network, based on the amount and consensus of evidences, and the localization of proteins (here and here)
• A shortcut to simply retrieve all curated ligand-receptor network resources almost without any modification (here)

[annajbott]

Yes is there a shortcut to the curated_ligand_receptor_interactions resources in the python version of Omnipath?

[deeenes]

Hi @annajbott - Not yet unfortunately. It would require translating this code to Python. I think it would make sense to do it in the current client, even though in the next 6-12 months both the client, the database and the server API will be replaced with completely new designs, this script will be easy to migrate to the new setup. I'm asking around in our group who might be interested doing it.

[annajbott]

Hi @deeenes thanks for your reply! Is any other further curation involved in the R script other than this? I.e. filtering for ligand and receptor pairs and using sources that are more trusted?

args = {'source': 'resource_specific', 'scope': 'generic'}
lig_rec = omnipath.interactions.import_intercell_network(
    include = ['omnipath', 'ligrecextra'],
    transmitter_params = {'parent': 'ligand'} | args,
    receiver_params = {'parent': 'receptor'} | args,
)
curated_resources = ['Guide2Pharma', 'HPMR', 'ICELLNET', 'Kirouac2010',
        'CellTalkDB', 'CellChatDB', 'connectomeDB2020', 'CellPhoneDB', 'Cellinker', 'talklr', 'SignaLink3']
pattern = '|'.join(curated_resources)
filtered_lig_rec = lig_rec[lig_rec['sources'].str.contains(pattern, case=False, na=False)].reset_index(drop=True)

Thanks!