comparison across conditions

[Wulin-Tan]

comparison across conditions

[hmbaghdassarian]

In the meantime, perhaps @earmingol or @hmbaghdassarian can provide a temporary function to convert liana's results into the format needed by tensor?

@earmingol has included a function in cell2cell to do that (tensor.external_scores.dataframes_to_tensor). You can run LIANA on each sample/context separately, put the outputs in a dictionary, then use the dataframes_to_tensor function to create the tensor.

[earmingol]

In the meantime, perhaps @earmingol or @hmbaghdassarian can provide a temporary function to convert liana's results into the format needed by tensor?

@earmingol has included a function in cell2cell to do that (tensor.external_scores.dataframes_to_tensor). You can run LIANA on each sample/context separately, put the outputs in a dictionary, then use the dataframes_to_tensor function to create the tensor.

To add more about this, once you build the tensor using the function that @hmbaghdassarian mentioned, you can easiy run Tensor-cell2cell using the function run_tensor_cell2cell_pipeline in this documentation

Tensor-cell2cell runs on Python, and can be found in this repo, so you can directly integrate liana-py and Tensor-cell2cell.

[dbdimitrov]

Hi @Wulin-Tan,

We're actually working on an extensive tutorial with LIANA and tensor, so this will come soon.

In the meantime, perhaps @earmingol or @hmbaghdassarian can provide a temporary function to convert liana's results into the format needed by tensor?

Daniel

[dbdimitrov]

Is it possible to add this feature (or even other features) to liana-py?

liana-py will be the development focus from now on, so all meaningful features from R will be also implemented here, along with certain extensions.

Daniel

[dbdimitrov]

Hi @Wulin-Tan,

With developments on both sides (i.e. with @earmingol and @hmbaghdassarian) we have now made a brief tutorial how to combine liana and tensor-cell2cell:
https://liana-py.readthedocs.io/en/latest/notebooks/liana_c2c.html

I'm closing the issue, but feel free to open others if you encounter issues.