Seurat object missing after running PCA

[FionaMoon]

Hi all, I've just updated my Seurat from V4 to V5 and found that my Seurat object was missing after RunPCA.

Here's my object before running PCA:

> obj
Assay (v5) data with 20204 features for 685 cells
Top 10 variable features:
 HBG2, HBG1, MPO, LYZ, HBB, S100A8, S100A9, PF4V1, RETN, CXCL8 
Layers:
 counts.sgRNA, counts.Control, data.sgRNA, data.Control, scale.data 

After running this code, my obj disappear.

 > obj <- RunPCA(obj, features = VariableFeatures(object = obj))
> obj
 A dimensional reduction object with key PC_ 
 Number of dimensions: 50 
 Number of cells: 685 
 Projected dimensional reduction calculated:  FALSE 
 Jackstraw run: FALSE 
 Computed using assay: RNA 

If I tried to run FindNeighbors & FindClusters, I'd get the error like this Error: Not an S4 object.

[samuel-marsh]

Hi,

Jot member of dev team but hopefully can be helpful. So your input to RunPCA isn’t a Seurat object but an assay object. So it appears somewhere before this there was issue.

Can you post full code you ran?

Best, Sam

[redtorrentCN]

**Hi, friends

I have the same problem. this is my code:**

`install.packages('Seurat')

library(Seurat) library(SeuratData) library(patchwork) library(dplyr) library(ggplot2)

ifnb.inte <- LoadData("ifnb") ifnb <- split(ifnb.inte[["RNA"]], f = ifnb.inte$stim) ifnb

ifnb <- NormalizeData(ifnb) ifnb <- FindVariableFeatures(ifnb) ifnb <- ScaleData(ifnb) ifnb <- RunPCA(ifnb)'**

this is my object before running PCA:

this is object after running:

in addition, this is data structure of ifnb.inte:

it's changed when I split by "stim", something like active.ident is gone, I don't know Whether it is normal:

p.s. I also maintain Seurat V4 in a different library. Above. thanks. @samuel-marsh

[FionaMoon]

Hi, My code almost seems like yours @redtorrentCN and I finally figured out what's wrong with this.

If I use obj <- split(obj, f = obj$Method) instead of obj[["RNA"]] <- split(obj[["RNA"]], f = obj$Method), then everything goes on well!

[redtorrentCN]

thanks, @FionaMoon , I believe this is useful, but I have solved this issue with reinstalling!
by the way, Have you encountered any of the following problems when runed scvi integration: obj<- IntegrateLayers( object = obj, method = scVIIntegration, new.reduction = "integrated.scvi", conda_env = "./anaconda3/envs/scvi-env", verbose = FALSE )

Error in object[][features, ] : incorrect number of dimensions In addition: Warning message: In LayerData.Assay5(object, layers = "counts") : multiple layers are identified by data.Smart-seq2 data.CEL-Seq2 data.10x_Chromium_v2_A data.10x_Chromium_v2_B data.10x_Chromium_v3 data.Drop-seq data.Seq-Well data.inDrops data.10x_Chromium_v2 only the first layer is used

Above. thanks for your answers.

[FionaMoon]

I'm afraid I can't answer this question. I run Seurat on Windows system which doesn't have conda environment.

---- Replied Message ---- | From | @.> | | Date | 01/03/2024 18:45 | | To | @.> | | Cc | @.>@.> | | Subject | Re: [satijalab/seurat] Seurat object missing after running PCA (Issue #8219) |

thanks, @FionaMoon , I believe this is useful, but I have solved this issue with reinstalling! by the way, Have you encountered any of the following problems when runed scvi integration: obj<- IntegrateLayers( object = obj, method = scVIIntegration, new.reduction = "integrated.scvi", conda_env = "./anaconda3/envs/scvi-env", verbose = FALSE )

Error in object[][features, ] : incorrect number of dimensions In addition: Warning message: In LayerData.Assay5(object, layers = "counts") : multiple layers are identified by data.Smart-seq2 data.CEL-Seq2 data.10x_Chromium_v2_A data.10x_Chromium_v2_B data.10x_Chromium_v3 data.Drop-seq data.Seq-Well data.inDrops data.10x_Chromium_v2 only the first layer is used

Above. thanks for your answers.

— Reply to this email directly, view it on GitHub, or unsubscribe. You are receiving this because you were mentioned.Message ID: @.***>

[redtorrentCN]

Fine, also thank you, bro.

[Gesmira]

Hi, Looks like the issue has been resolved in the comments. Closing now, but please feel free to reopen if the issue persists. Thanks!