Muon: multiple set of data comparative analysis

Hi ,
iā€™m using Muon for CITE-Seq data analysis. Can you help me to resolve certain issues :slight_smile:

  1. I need to perform comparative analysis of multiple groups in same notebook. How to INTEGRATE or analyze multiple sets at a time ?
  2. How can i do comparative Differential expression , KEGG and Trajectory analysis using Muon ?
    Please, let me know.
    Best regards,
    Babu ( md.babu.mia@mssm.edu)

Hey!

If this is a question of multiple batches and the epitopes are the same in those batches, there are various options of accounting for the batch covariate. In muon, for instance, MOFA has a respective argument (see more about the groups_label here). TotalVI from scvi-tools uses the batch_key provided when setting up the AnnData (might still require to use AnnData to train this model at the moment). Overall, this is a rough direction of approaching CITE-seq analysis with multiple data batches.
If the epitopes are different, one has to figure out what integration strategy should be appropriate. For instance, aligning in the RNA space might be good enough in some cases. Training a MOFA model with using missing values for modality-specific epitopes can also be an option.

Differential expression, pathway analysis, trajectory analysis are enabled by other packages, e.g. see some of the scverse ecosystem packages here.
Just for a few pointers, in Python there are

1 Like