Conversion of sfe Object to AnnData

[aasingh2]

Hello,

I wanted to ask if there is a built-in function to convert a sfe object to AnnData object, or if there is a way to save the sfe object as a h5ad file to facilitate interoperability with Python.

Thanks!

[alikhuseynov]

I don't think it is planned right now, but maybe @lambdamoses has some ideas.
FYI, there is VoyagerPy and updated python workflow for Xenium
There will hopefully be integration with SpatialData pachterlab/voyagerpy#29

[lambdamoses]

For now you can use some package that converts between SCE and AnnData such as zellkonverter because SFE inherits from SCE, and then deal with the spatial part separately in a manner that is consistent with squidpy. I think it would be a valuable new feature to incorporate.

Also, the ArtifactDB project has the alabaster packages for language agnostic serialization of Bioc objects from R; the Python counterpart are the dolomite packages. I just submitted alabaster.sfe to Bioc recently; there's no dolomite for SFE yet, but I should we should write it.

[vraghu90]

On the same lines, I have an AnnData object which I would like to convert to spatial feature experiment. Is there a workflow to do this and and run the Moran's I calculation?

[lambdamoses]

Not yet, but I think there should be. For now, you can convert AnnData to SCE with something like zellkonverter first and then add the geometries and coordinates. Please consider opening a pull request.

[lambdamoses]

Also, you don't need to convert AnnData to SFE to compute Moran' I. Squidpy can do that, so can voyagerpy: https://github.com/pachterlab/voyagerpy/tree/dev2 Although at present only the R implementation of Voyager has the bivariate and multivariate spatial statistics.

[vraghu90]

Can we do a heatmap of the moran's I scores using voyagerpy? Best, Vineeth

…

--- *Vineeth Raghuram, Ph.D.* *(603)-943-0489*

On Thu, Nov 14, 2024 at 2:28 PM Lambda Moses ***@***.***> wrote: Also, you don't need to convert AnnData to SFE to compute Moran' I. Squidpy can do that, so can voyagerpy: https://github.com/pachterlab/voyagerpy/tree/dev2 Although at present only the R implementation of Voyager has the bivariate and multivariate spatial statistics. — Reply to this email directly, view it on GitHub <#56 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/BAT7LOKQOMKIOLAPQYKINU32AT2WHAVCNFSM6AAAAABRYTMEY6VHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDINZXGI2DEMBUGU> . You are receiving this because you commented.Message ID: ***@***.***>

[lambdamoses]

What kind of heatmap? Do you mean plotting local Moran's I's at spatial locations or global Moran's I across genes and samples?

[vraghu90]

Both actually.

…

--- *Vineeth Raghuram, Ph.D.* *(603)-943-0489*

On Thu, Nov 14, 2024 at 10:18 PM Lambda Moses ***@***.***> wrote: What kind of heatmap? Do you mean plotting local Moran's I's at spatial locations or global Moran's I across genes and samples? — Reply to this email directly, view it on GitHub <#56 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/BAT7LOPH3XY5PPA6I7PUE7D2AVRXDAVCNFSM6AAAAABRYTMEY6VHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDINZXHA3TEMJTHE> . You are receiving this because you commented.Message ID: ***@***.***>

[lambdamoses]

You can plot local Moran's I with the plotLocalResult function in Voyager (I think it's snake case in the Python implementation) but you need to do some data wrangling yourself to plot Moran's I across genes and samples. In the R implementation, global spatial results for gene expression are stored in rowData and the column names indicate sample_id. rowData is an S4 DataFrame; you can extract the relevant columns and convert that into a matrix which can be plotted as a heatmap with the pheatmap package. I'm not sure how well the Python implementation of Voyager handles different samples though.

[lambdamoses]

Any more questions? If not then I'll close the issue.

[vraghu90]

Quick question on this. Is there a workflow where I can convert a previously analyzed/processed seurat object(Xenium) into a SFE to perform Voyager operations?

[alikhuseynov]

Quick question on this. Is there a workflow where I can convert a previously analyzed/processed seurat object(Xenium) into a SFE to perform Voyager operations?

Yes, take a look here -> https://pachterlab.github.io/SpatialFeatureExperiment/dev/articles/SFE.html#coercion-from-seurat