Cell annotation using integratelayers() and SCTransform()

[michaelhojungyoon]

I'd like to start off by stating that I'm rather new to analyzing single-cell data and am very thankful for the authors for developing this package.

Currently, I have a list of Seurat objects that I wish to annotate using anchors as part of the "Mapping and annotating query datasets" workflow. However, I have another Seurat object containing HLCA data. To keep dimensionality reduction consistent, I attempted to split() based on sample type, run SCTransform(), and IntegrateLayers() for rpca which all passed for the samples, but for the reference Seurat object, there is no sample column on the metadata slot to split with and IntegrateLayers() fails.

If anyone has a better workflow or solution, please let me know.

[alexwskh]

So you have two seurat objects, one is your single cell experiment and the other is the lung cell atlas? You loaded them separately and then merged them together, before trying to split and trying to integrate? You can create a "sample" metadata before the merge and it will be brought over... Like,

seurHLCA$sample <- "HCLA"
seurMyExperiment$sample <- "MyExperiment"

Do your merge,

Then you can split(seur_merged, f =seur_merged$sample)

That should give you two separate counts layers that it will want for SCTransformation (I think, not an expert)...