Question about adata_st.uns['transcripts'] format

[MarleneLi]

Hi, thank you for developing such a helpful tool! I tried to follow your tutorial but encountered some problems with adata_st.uns['spatial'].

According to the official 10x Genomics Xenium documentation, the transcripts.parquet file has the format shown in the attached screenshot (Fig. 1). However, in your tutorial, adata_st.uns['transcripts'] uses a different column set and naming convention — specifically ['gene', 'uID', 'absX', 'absY'].

I’m trying to understand the mapping between these two formats:

• Should uID correspond to transcript_id or cell_id from the Xenium file?
• I initially set uID to transcript_id, but this caused an error.
• In my dataset, all cell_id values are "UNASSIGNED" (see Fig. 2). How should this case be handled when preparing adata_st.uns['transcripts']?

Any clarification on the conversion process from the official Xenium transcript format to the tutorial’s expected format would be very helpful.

Thanks!

[bhavaygg]

Hi,

Thank you for using the tool. We do use a different format than the Xenium format but it is very similar. The uID corresponds to the cell_id. You would need to remove the UNASSIGNED transcripts. These are the transcripts sequenced outside any cell. Additionally, the feature_name corresponds to the gene name and this will also have redundant genes added for quality control. These do generally start with prefixes like NegControl among others, so please check for these and remove them as well.

[MarleneLi]

Thank you so much for your help and clarification! I'll try your suggestions.