Script for doublet finder with seurat 5

[synatkeamsk]

Dear Chris and team,

Sorry for asking you and your team again. Hope you do not mind. You said Doublet Finder should work with Seurat 5 now. I am using Seurat 5.0.1. I used the following scripts!

#read data of data !
data<- Read10X(data.dir = "patient_1")

#create seurat object
data<- CreateSeuratObject(counts = data, project = "164S", min.cells = 3, min.features = 200)

#SCTransform !
data<- SCTransform(data, method = "glmGamPoi", vars.to.regress = "percent.mt", verbose = FALSE)
data <- RunPCA(data, seed.use = 12345,verbose = FALSE)

#PC plot
ElbowPlot(data, ndims = 50,reduction = "pca") +
  ggtitle("Number of principle component")  + 
  theme_minimal() +
  theme(plot.title = element_text(hjust = 0.5))

#Run umap 
data <- RunUMAP(data, seed.use = 12345, dims = 1:45)
DimPlot(data, reduction = "umap", label = TRUE, label.size = 4) +
  theme_void() +
    NoLegend() 

library(DoubletFinder)

### ## pK Identification
doublet.data<- paramSweep_v3(data, PCs = 1:45, sct = TRUE)

 **Error in paramSweep_v3(doublet.data, PCs = 1:45, sct = TRUE) : 
  no slot of name "counts" for this object of class "Assay5"**

doublet.stats_data<- summarizeSweep(data, GT = FALSE)
bcmvn_data <- find.pK(doublet.stats_data)

Since chris has tested and has shown to work with seurat V5. I am wondering whether you could have a look at my script and suggested why it generated error. Hope you could help!

Kind Regards,

Synat,

[synatkeamsk]

Apologise. I did not update the DoubletFinder package. After removing old package and install new one. All worked now. Thanks you, Chris and team for your wonderful work!

Kind Regards,

synat

[olneykimberly]

I'm getting the same error. I removed the old packages and updated but I still have the same issue.

[synatkeamsk]

I think the function have changed. From your UMAP object, you can try the following script

## pK Identification
doublet.A184S2<- paramSweep(Normalized.184S2, PCs = 1:40, sct = TRUE)
doublet.stats_A184S2 <- summarizeSweep(doublet.A184S2, GT = FALSE)
bcmvn_A184S2 <- find.pK(doublet.stats_A184S2)

#save bcmv_A164S object
saveRDS(bcmvn_A184S2, file = "bcmvn_A184S2.RDS")

#ggplot 
ggplot(bcmvn_A184S2, aes(pK, BCmetric, group = 1)) +
  geom_point() +
  geom_line() +
  theme_minimal()

pK <- bcmvn_A184S2 %>% # select the pK that corresponds to max bcmvn to optimize doublet detection
  filter(BCmetric == max(BCmetric)) %>%
  select(pK) 
pK <- as.numeric(as.character(pK[[1]]))

## Homotypic Doublet Proportion Estimate 
annotations <- [email protected]$seurat_clusters
homotypic.prop <- modelHomotypic(annotations)           
nExp_poi <- round(0.025*nrow([email protected]))  ## Assuming 2.5 percent doublet
nExp_poi.adj <- round(nExp_poi*(1-homotypic.prop))

# run doubletFinder 
Normalized.184S2 <- doubletFinder(Normalized.184S2, 
                                     PCs = 1:40, 
                                     pN = 0.25, 
                                     pK = pK, 
                                     nExp = nExp_poi.adj,
                                     reuse.pANN = FALSE, sct = TRUE)

# visualize doublets
view([email protected])
DimPlot(Normalized.184S2, reduction = 'umap', group.by = "DF.classifications_0.25_0.14_89") +
  theme_minimal() +
  theme(plot.title = element_text(hjust = 0.5))

Please use SCT= TRUE if you use SCTransform or SCT= FALSE if you use standard seurat pipeline

[olneykimberly]

@synatkeamsk Thank you!!! I followed it as you have described and it works now as expected. Thank you so much for the help and fast reply.

[Yijia-Jiang]

Hi, I am facing the same issue, could you let me know which version did you use? Thank you so much!

[synatkeamsk]

@Yijia-Jiang,

Here is my package version packageVersion("Seurat") [1] ‘5.0.1’

packageVersion("DoubletFinder") [1] ‘2.0.4’

Hope you could fix it. @olneykimberly may be able to comment further.

Thanks,

[Yijia-Jiang]

@synatkeamsk Thank you very much for such a quick response! I am using the exact same version, but got an error when running paramSweep function, and the error shows "Error in .subscript.2ary(x, , j, drop = TRUE)". I wonder if you have ever encountered that before?

[synatkeamsk]

I have not encountered this error before. I have just re-run my script again. I did not see any error! These are the sort scripts I run for one sample.

#read data of 164S !
A164S<- Read10X(data.dir = "164S")

#create seurat object
A164S<- CreateSeuratObject(counts = A164S, project = "164S", min.cells = 3, min.features = 200)
A164S

#SCTransform !
Normalized.164s<- SCTransform(A164S, method = "glmGamPoi", verbose = FALSE)
Normalized.164s <- RunPCA(Normalized.164s, seed.use = 12345,verbose = FALSE)

ElbowPlot(Normalized.164s, ndims = 50,reduction = "pca") +
  ggtitle("Number of principle component")  + 
  theme_minimal() +
  theme(plot.title = element_text(hjust = 0.5))

#Run umap 
Normalized.164s <- RunUMAP(Normalized.164s, seed.use = 12345, dims = 1:40)
DimPlot(Normalized.164s, reduction = "umap", label = TRUE, label.size = 4) +
  theme_void() +
    NoLegend() 

library(DoubletFinder)

## pK Identification
doublet.A164S<- paramSweep(Normalized.164s, PCs = 1:40, sct = TRUE)

I did not see any error until this step, but I understand your data may be different from mine.

[olneykimberly]

I am using the same versions of Seurat. For doubletFinder, I copied the git repo and sourced the functions. Like so: source("/tools/DoubletFinder/R/parallel_paramSweep.R")