Aggregate samples into single object

nf-core / spatialvi

Pipeline for processing spatially-resolved gene counts with spatial coordinates and image data. Designed for 10x Genomics Visium transcriptomics.

https://nf-co.re/spatialvi

MIT License

48 stars 17 forks source link

Aggregate samples into single object #56

Open grst opened 1 year ago

grst commented 1 year ago

Description of feature

Would be nice to have a single object instead of inidividual h5ad files. Optional batch correction for integration.

grst commented 1 year ago

I would like to refactor a bit how AnnData objects are created and handled

instead of having a separate AnnData object after each step, I'd just go for two: one which has low quality spots removed and one that doesn't. Raw counts and normalized values can be stored in a single object using adata.layers.
rely on scanpy or squidpy to read visium data (@fasterius, what was again the motivation to have a custom python script for that in the first place?)
merge all objects into a single AnnData object at the end, with optional export to a SpatialData object (#39).

I'd be happy to put a PR together.

@fasterius / @cavenel WDYT?

fasterius commented 1 year ago

Sounds good to me! I haven't really done any of the coding inside the reports, so don't have much to add - you'll have to ask @cavenel regarding your second point about the custom script. Might be due to legacy code from the Jackson Laboratory version of the pipeline, but that's just a guess.

fasterius commented 3 months ago

With the latest updates to the pipeline I think that we fulfill your first two points @grst. Regarding the third point, do you mean "merge" as in some kind of integration or just something else?

grst commented 3 months ago

I just mean storing all samples in a single SpatialData object rather than having one SpatialData object per sample.