Question related to normalization about spotlight HOT 3 CLOSED

Hi @JoyOtten
Thank you very much for using SPOTlight.
To identify the differentially expressed genes we use SCTransform as it is the recommended workflow in Seurat to normalize the data. SPOTlight in turn uses unit-variance normalization for both the sc and spatial count matrices.
Hope this helps and please don't hesitate to get back to me if you have any other questions.

from spotlight.

Hi @MarcElosua
I am also confused about a similar question.
I understand that you use SCTransform to follow the recommended workflow to identify DE genes.
However, in the example of deconvolution, you suggest as follows

spotlight_ls <- spotlight_deconvolution(se_sc = cortex_sc,
                                        counts_spatial = anterior@assays$Spatial@counts,
                                        clust_vr = "subclass",
                                        cluster_markers = cluster_markers_all,
                                        cl_n = 50,
                                        hvg = 3000,
                                        ntop = NULL,
                                        transf = "uv",
                                        method = "nsNMF",
                                        min_cont = 0.09)

You used the assay of 'SCT' (spatial) for the counts_spatial while 'RNA' for the se_sc (I recognized default assay from the documentation for spotlight_deconvolution)?

As I known, the slot of counts are different between 'SCT' and 'RNA'. Why don't you use the same assay or I missed something?

from spotlight.

hi @quanc1989,
As I mentioned in the previous comment SPOTlight uses unit variance normalization to run the model. Regarding the assay of choice to get the DE it is flexible and does not need to be SCT!
Hope this answers your question.

from spotlight.