multi sample import normalization and plot customisation.

[rLannes]

Hi, First thank you for providing such a package to the community.

I want to use metacells to analyses multi-sample (different patients, different libraries, all 10X) I wonder if I should apply a normalization step before importing the matrix?

I would like to know can I color cells based on samples of origins? For what I have seen in the documentation cells can only be colored accordingly to metacells or genes expressions.

For the cell/metacells/graph plot is there a way to access the layout: cells position, metacells position and link? So it is easier to customize and tweak?

[orenbenkiki]

Hi,

Sorry for the late response. We are finalizing the paper and the package so do not use it for real-world analysis yet.

To your questions:

• No, you shouldn't normalize the data, we deal with this internally in the package (mainly by downsampling).
• If you run the umap computation functions, the coordinates are stored as per-observation umap_x, umap_y annotations in the metacells annotated data object. Likewise theres a per-observation-pair obs_outgoing_weights sparse matrix containing the graph used for UMAP. Note this graph and projection are only for visualization and should not be used for actual analysis.