Same data, different Pearson Residuals

[luxhexmaster]

Hi all,

Thank you so much for all the work you've put into the Seurat pipeline!

I'm re-running a dataset that I ran back in March of 2024, but I'm obtaining different UMAP clustering despite using the same Seurat version (5.0.0) and the same script. Using the old annotated object I saved from my previous run as a comparison, I find that the only difference is in the scale.data of SCT which should store the Pearson residuals. Is there any reason I would get different Pearson residuals if the 'RNA' data is the same?

I checked and both objects were run with the same versions of Seurat, SCTransform, SeuratData and SeuratObjects.

Any help would be greatly appreciated. Thanks!

[saketkc]

Hi @luxhexmaster, can you provide a bit more context (code/objects/environment) for me to be able to debug?

[luxhexmaster]

Hi Dr. Choudhary, just sent you the scripts and objects via email. Thanks!