Cannot use images from object converted from .h5ad to .h5seurat [Spatial Transcriptomics]

[akstks]

Hello dear Seurat developers,

I have been using Seurat for a while now analyzing scRNA-seq data, and wanted to try out some spatial analysis to familiarize myself with that topic of transcriptomics.

I found this publicly available dataset that has 4 samples (A1, B1, C1, D1) that have been processed with Python pipelines:

https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE171351

Sadly, they do not provide the raw data to create a Seurat object from scratch, so I converted the .h5ad provided object to an .h5Seurat object as instructed through the Seurat Disk package.

Loading the object with LoadH5Seurat produced a Large Seurat object spatial <- LoadH5Seurat(GSE171351_combined_visium.h5seurat)

however: spatial@images is empty and what appears to be the image embeddings are found in spatial@reductions[["spatial"]]@misc

Here is a picture of the RStudio view panel to get a better look:

Since these appear to be already analyzed data, I wanted to just look at SpatialFeaturePlots for various genes, but I cannot seem to figure out how to do that, without the images being in the "correct" place, or of class "VisiumV1".

Do you have any ideas on how to proceed? I am really new to the spatial aspect so any help would be much appreciated!

[hbzxm]

@akstks Did you find the solution?

[akstks]

@hbzxm No sadly I did not. I gave up.

[cellphone12]

It seems to record the location of each cell in meta.data.

[cellphone12]

It seems to record the location of each cell in meta.data.

[rsatija]

Apologies that we cannot support this. The complexity of spatial data types is too great that we cannot support generic conversion of spatial h5ad to seurat files. You can recreate a Seurat object from the spatial transcriptomic datasets in order to utilize the Seurat-spatial toolkit

[Carissimo]

Did you resolve this issue? I have the same problem with stereoseq spatial data