Interpreting DEG Differences Between 8×8 Bins and Cell-Segmented VisiumHD Tumor Data

[dianegg25-tech]

Hi,

first of all, thank you for the great work on this tool!

I’m working with the new VisiumHD data with cell segmentation following the instructions on the pull request 10028, and I noticed something I’d like to better understand.

When I compare differentially expressed genes (DEGs) between 8×8 bins and cell-segmented bins for the same cell type (same spatial location) , I see that some genes present in the 8×8 DEG results are “lost” in the cell-segmentation results. More specifically, in the 8×8 bins my tumor cells show expression of markers typically associated with normal cells, but these markers disappear when I use the cell-segmented bins, having just markers of tumoral cells. I am using the same QC, normalization step and the same resolution for both of them.

I am quite new in all the transcriptomic analysis part, so I would like to know:

• Why these markers might appear in the 8×8-bin analysis but not in the cell-segmented analysis?
• Which approach is considered more biologically reliable for downstream interpretation—using the 8×8 bins or the cell-segmented data?

Thanks in advance!