How to re-label cell identity after the analysis was already done?

[ilcink]

How to change a cell identity label without re-analysing the whole dataset?

I did
cellchat <- setIdent(cellchat, ident.use = "cell_type") # set "cell_type" as default cell identity (the new ones)
levels(cellchat@idents)
groupSize <- as.numeric(table(cellchat@idents))

it shows the right new idents, but the plots still show the initial cell annotation.

Thanks!

[sqjin]

Hi @natalkon , I just updated the package to support such action. You can re-label cells without re-doing computeCommunProb

[ilcink]

Dear @sqjin ,

I tried to update the package simply by
devtools::install_github("sqjin/CellChat")

But I got
Error: Failed to install 'CellChat' from GitHub:
(converted from warning) installation of package ‘backports’ had non-zero exit status

How can I overcome it?

Thanks!

[ilcink]

Dear @sqjin ,
I have manually deleted the "backports" folder from the library, downloaded the package "backports_1.1.9.zip" from CRAN https://cran.r-project.org/web/packages/backports/index.html and placed the new "backports" into the library.
After that updating the package with
devtools::install_github("sqjin/NMF")
devtools::install_github("sqjin/CellChat")
was no problem.

But after
identity = data.frame(group = sc_data$cell_type, row.names = names(sc_data$cell_type)) # create a dataframe consisting of the cell labels
unique(identity$group)
cellchat <- addMeta(cellchat, meta = identity, meta.name = "cell_type")
cellchat <- setIdent(cellchat, ident.use = "cell_type") # set "cell_type" as default cell identity
levels(cellchat@idents)
groupSize <- as.numeric(table(cellchat@idents))

Re-labelling worked for
CellChat::netVisual_aggregate(cellchat, signaling = c("TGFb"), layout = "circle", vertex.size = groupSize)

but still did not happen in
CellChat::netAnalysis_river(cellchat, pattern = "outgoing")
or in
CellChat::netAnalysis_dot(cellchat, pattern = "outgoing")

Do you know what could be the reason for that?

Thanks!

[sqjin]

@natalkon As I mentioned that "You can re-label cells without re-doing computeCommunProb", thus you need to re-run all the functions after computeCommunProb.

[ilcink]

@sqjin That is amazing! Thank you!!!

[ruohuchengxhe]

@sqjin
Hi, I tried to change one label in both the metadata and idents:

levels(cellchat@meta$labels)[levels(cellchat@meta$labels) =='Mast cells' ] <- 'Mc 4'
levels(cellchat@idents)[levels(cellchat@idents) =='Mast cells' ] <- 'Mc 4'
table(cellchat@idents)

The label was changed successfuly.

And then I run the code after computeCommunProb, but when I try to plot, the label is still 'Mast cells', even for the CellChat::netVisual_aggregate function.

Could you please give advice?
Thanks!

[sqjin]

Hi @ruohuchengxhe , you should do it levels(cellchat@meta$labels)[levels(cellchat@meta$labels) =='Mast cells' ] <- 'Mc 4' cellchat <- setIdent(cellchat, ident.use = "cell_type") # set "cell_type" as default cell identity

[ruohuchengxhe]

Hi @ruohuchengxhe , you should do it levels(cellchat@meta$labels)[levels(cellchat@meta$labels) =='Mast cells' ] <- 'Mc 4' cellchat <- setIdent(cellchat, ident.use = "cell_type") # set "cell_type" as default cell identity

Sure! Didn't do this. Thanks!