Seurat: Some detail issues and suggestions in Seurat v3

[QiangShiPKU]

Hi,
I have found some issues in Seurat v3, and I want to provide some suggestions to enhance this version.

1. In the PBMC 3k tutorial, we could not find function VizDimReduction. Should it be replaced by
   VizDimLoading?

2. In the PBMC 3k tutorial, we could not find argument thresh.use for FindMarkers or FindAllMarkers function. Should it be replaced by logfc.threshold?

3. For RidgePlot, the argument log says "plot Y axis on log scale", but I found it seems "plot X axis on log scale" (X axis is expression).

4. For RidgePlot, VlnPlot and so on, I don't understand the function of argument split.by.

5. For ScaleData, arugments do.par and num.cores are deprecated. Why? Does this allow parallel processing?

6. Last but not least #810
Seurat v3.0 can transfer information from reference to query datasets. It is great!

I think it is very useful to project the query dataset onto the same UMAP layout produced by reference.
This link may help, https://cran.r-project.org/web/packages/umap/vignettes/umap.html.

I have realize this fuction on basic of Seurat v2.3.4 without much challenge, but I cannot integrate this into original package well. I think it is easy for you to hand it.

Thanks !

[andrewwbutler]

Hi,

1. Yes, that is correct. We've renamed VizDimReduction to VizDimLoadings to make it more apparent what is being plotted.
2. Yes, the thresh.use parameters has been replaced by logfc.threshold, again for clarity.
3. Yes, the x-axis scale is what is being logged here. The documentation has been updated accordingly.
4. The split.by parameter allows you to split the idents by some categorical variable (for example to visualize differences between treated/untreated groups within a celltype). Should provide a similar visualization to the plots seen in this post.
5. In v3, we've moved to the future framework for parallelization. There will be more demonstration examples of this with the full release but essentially, for the user this will amount to setting the plan.
6. That is a nice feature of UMAP and is something we could consider adding once some bandwidth frees up.