Skip to content

Latest commit

 

History

History
64 lines (34 loc) · 2.63 KB

FAQ.md

File metadata and controls

64 lines (34 loc) · 2.63 KB

FAQ - Frequently Asked Questions

1. The 2D UMAP looks different after running Gruffi. How can I keep the original UMAP?

The SetupReductionsNtoKdimensions() function simply calls the RunUMAP() function, and backs it up into obj@misc$reductions.backup , so if you provide the exact same parameters, it should reproduce the UMAP you generated earlier.

An alternatively you can

# 1. backup your old 2D umap
combined.obj@misc$reductions.backup$umap2d <- combined.obj@reductions$umap


# 2. calculate and backup the new 3D umap | "dimensions=3"
combined.obj <- Seurat.utils::SetupReductionsNtoKdimensions(obj = combined.obj, nPCs = 50, dimensions=3, reduction="umap")

# 3. Recall the old 2D umap from @misc
combined.obj@reductions$umap <- combined.obj@misc$reductions.backup$umap2d
# The 3D is in ...reductions.backup$umap3d


# 4. Gruffi will find the 3D umap in @misc - it is needed for the reclassification step.

2. How to use Gruffi with different integrations?

Gruffi assumes that you used Seurat, thus granule clustering runs on assays@RNA or assays@integrated.

Typically, integration methods simply provide an alternative reduction to PCA such as iNMF for LIGER batch correction. Simply provide this reduction when calculateing the 3D UMAP (see 2.).

If you integrated with a method that created a different assay (e.g.: @BlaBla), provide the assay name in AutoFindGranuleResolution( , assay = "BlaBla").

3. How to use Gruffi when providing a custom set of genes, instead of a GO-term?

Gruffi can work with any set of genes to classify cells. In the current implementation you can provide 2 positive and 2 negative filtering terms (in the default implementation we use 2 positive and 1 negative terms).

# You add scores of your gene set and calculate granule averages:
heGENES <- c("TMSB4X", "NRXN3", "SNTG1","SOX4", "TUBA1A", "NRXN1", "TMSB10", "ACTG1", "ROBO2","ACTB")
combined <- AddCustomScore(obj = combined.obj, genes = heGENES)
# Scorename: "Score.heGENES"

# Then calculate granule averages
combined.obj <- CustomScoreEvaluation(obj = combined.obj, custom.score.name = "Score.heGENES")

Finally continue with the Gruffi pipeline as normal, by calling the Shiny app. This extension of the Gruffi workflow is less tested, so if you find bugs / encounter an error, please let us know.

4. Does it work with Seurat v5+?

Yes, it should. Please raise an issue if you experience problems.

5. Does it install and run work with R v4.4.X?

Yes, it should. You may need to manually install terra and rgl dependencies. Please raise an issue if you experience problems.