Equivalent Seurat objects saved as .rds have different hashes

[falexwolf]

DataFrame serialization behaves as it should.

But it seems that saveRDS() or something in Seurat mixes in a timestamp so that the hash becomes meaningless and one ends up with this messy situation upon re-running a notebook:

https://lamin.ai/laminlabs/lamindata/artifacts?filter[and][0][or][0][visibility][eq]=1&filter[and][1][or][0][is_latest][eq]=true&filter[and][2][or][0][transform.name][eq]=analysis_renal_cell_carcinoma.Rmd

We should look into what saveRDS() is doing and for the time being, use the key argument when creating an artifact from_path(..., key="analysis/my_dataset") instead of the description. Runs with the same key will at least put artifacts in the same version family.

Ahhh, this is messy. People won't like this. 😔