search

saezlab / liana-py

LIANA+: an all-in-one framework for cell-cell communication
http://liana-py.readthedocs.io/
GNU General Public License v3.0
134 stars 15 forks source link

The mouse resource are missing important interactions that should be possible to convert #76

Closed johan-gson closed 1 month ago

johan-gson commented 7 months ago

Describe the bug Hi and thanks for a nice tool. For example, Cxcl9 is missing as ligand in the database, while Cxcl10 is there, and they have many receptors in common. Converting CXCL9 to Cxcl9 should be doable? I'm guessing something has gone wrong and that many interactions are missing in the mouse data.

To Reproduce Just look in the database

dbdimitrov commented 7 months ago

Hi @johan-gson,

Indeed, there are some interactions missing in the mouse data, very likely as their orthologues were missing from the mapping reference (homologene), that we used.

I plan to rebuilt the Consensus and MouseConsensus resources soon using orthologues from UniProt, and also provide a short tutorial for resource conversion.

I will keep you updated.

johan-gson commented 7 months ago

Thank you!

Best,

Johan

dbdimitrov commented 1 month ago

Hi @johan-gson, sorry for the delay, but I finally managed to implement orthology conversion that I'm happy with. Note the min_evidence and one_to_many parameters, as depending on how strict you want to be with the prior and the mapping, might have a notable effect on the conversion.

https://liana-py.readthedocs.io/en/latest/notebooks/prior_knowledge.html#Obtain-Mouse-Homologs

johan-gson commented 1 month ago

Thanks, will have a look next time we need it! I'm sure many will find this most useful!