Ensembl IDs for LIANA inference?

[bhavyaac]

Hello, thank you for your wonderful work on this tool and corresponding publication.

I am looking to analyze scRNA-seq data with genes listed as Ensembl transcript IDs. If possible, I would like to avoid converting to gene symbols for the cell-cell interaction inference.

I have noticed that certain tools e.g. CellphoneDB can be run with Ensembl IDs. Is there any way to run LIANA directly with these IDs? Is there a certain set of databases and tools that I could specify in LIANA in order to allow this?

Please let me know what you think. Thank you!

[deeenes]

Hello,

Because the available database knowledge uses UniProt IDs and Gene Symbols, you have to convert these to Enselmbl Genome Transcript IDs. It will come with a loss of information due to all the intricacies of this ID conversion. The same why, I assume, you afraid of converting the IDs in your dataset to Gene Symbol. It would be interesting to compare LIANA results with the both kind of conversions. And I am wondering if people commonly do it this way.

For ID conversion I can recommend this function . @dbdimitrov will let you know how the converted database knowledge can be channeled into the LIANA pipeline, and also if I am wrong in anything above, I am not too familiar with LIANA.

Best,

Denes

[dbdimitrov]

Hi,

One can pass any custom resource as follows:

liana_res <- liana_wrap(testdata,
                        resource = 'custom', # resource has to be set to 'custom' to work with external resources
                        external_resource = ortholog_resource, # provide orthologous resource
                        method=[c](https://rdrr.io/r/base/c.html)('sca', 'natmi') # run only with sca and natmi for comp. time
                        )

e.g. see this tutorial : https://saezlab.github.io/liana/articles/liana_ortho.html

LIANA would work with any resource that has source_genesymbol and target_genesymbol as columns.