FindTransferAnchors with an integrated reference - what reduction to use?

[agreco92]

Hi, I'm running several pipelines in which I need to map query datasets onto a reference. In the Multimodal reference mapping tutorial, a reference integrated using CCA is used to predict cell type labels in a query dataset.

The ref.pancreas object has a non-integrated dimensionality reduction (pca) and an integrated one (cca). The FindTransferAnchors function is invoked with the argument reference.reduction = "pca" rather than cca. Is this correct? My intuitition suggests that the pca reduction will contain batch structure that confounds the mapping.

There could be some more subtle processing using the split layers but it's not easy to understand from the tutorial itself, so I'd be happy to receive some straightforward answer. Thank you for your help!