Error in blockwiseIndividualTOMs

[yt723]

Hello,

I can't solve the error below. Can you suggest what I should try to troubleshoot? Thank you so much.

Error in blockwiseIndividualTOMs(multiExpr = multiExpr, checkMissingData = checkMissingData, : Standard deviation of some genes is zero.

[smorabit]

Hi,

Could you please provide a bit more information in order for me to help you? Can you send all of the hdWGCNA code that you used? More information about your dataset could also help.

[yt723]

Hello,

I'm running the ConstructNetwork function below and facing the problem. My samples are PBMC and data was obtained by scRNAseq+CITEseq (BD Rhapsody). I integrated data from different libraries using Seurat's CCA method. After obtaining major cell types (such as B cells, CD8T cells, NK cells etc.), I did targeted clustering. Then I'm running WGNCA for those major cell types with seurat_clusters. The codes I'm using are working for most of the cell types, but I can't analyze one of the cell types because of the error message.

My codes: CD4Tm <- ConstructNetwork( CD4Tm, soft_power = 3, setDatExpr = FALSE, deepSplit = 2, overwrite_tom = TRUE, TOMType = 'unsigned', minModuleSize = 10, mergeCutHeight = 0.1, tom_name = 'CAV' )

Response: Calculating consensus modules and module eigengenes block-wise from all genes Calculating topological overlaps block-wise from all genes Flagging genes and samples with too many missing values... ..step 1 TOM calculation: adjacency.. ..will not use multithreading. Fraction of slow calculations: 0.804846 TOM: exit because 'adjacency' reported an error. Error in blockwiseIndividualTOMs(multiExpr = multiExpr, checkMissingData = checkMissingData, : Standard deviation of some genes is zero.

---

Default assay is set for integrated RNA. I chose genes by 'variable' for SetupForWGCNA. There are 477 genes selected. When I encountered the same issue, I changed 'max_shared' value in the MetacellsByGroups function and that sometimes solved the problem. But for the cell type above, it didn't work. I also tried to change SetupForWGCNA parameter from 'variable' to 'fraction', but this didn't work because of the error message (below).

CD4Tm <- SetupForWGCNA(CD4Tm, gene_select = "fraction", fraction=0.05, wgcna_name = 'wgcna_CAV') Error in cut.default(1:nrow(expr_mat), n_chunks) : invalid number of intervals

What else should I try? Thank you so much for your help.

Yuko

[smorabit]

Can you also show the code that you ran before ConstructNetwork?

Default assay is set for integrated RNA. I chose genes by 'variable' for SetupForWGCNA. There are 477 genes selected.

This seems like a very low number of genes, not sure if this is causing this particular issue but usually there are a few thousand genes included for network analysis. I actually have not tried running hdWGCNA with this few genes.

[yt723]

Hello,

We did targeted RNA sequencing and the number of genes included was less than 500. I have other seurat objects from the same experiment (same genes) and WGCNA worked for them. Only a few seurat objects are causing errors. The codes I ran before ConstructNetwork are below.

DefaultAssay(CD4Tm) <- 'integrated' CD4Tm <- SetupForWGCNA(CD4Tm, gene_select = 'variable', wgcna_name = 'wgcna_CAV') CD4Tm <- MetacellsByGroups( seurat_obj = CD4Tm, group.by = c("seurat_clusters", "SampleType"), k=25, max_shared = 10, assay = 'integrated', slot = 'data', ident.group = 'SampleType') CD4Tm <- NormalizeMetacells(CD4Tm) CD4Tm <- SetDatExpr(CD4Tm, group_name = "CAV", group.by = 'SampleType', assay = 'integrated', slot = 'data')

Thank you so much.

Yuko

[smorabit]

How many cells are there in the group that is giving you this error message?

Looking at your error message:

Error in blockwiseIndividualTOMs(multiExpr = multiExpr, checkMissingData = checkMissingData, : Standard deviation of some genes is zero.

I think that there's something wrong with the some of genes that you're using as input. However I am somewhat confused because SetDatExpr is supposed to look for genes with zero variance and remove them from the analysis. Maybe try selecting genes a different way. Unfortunately it is very difficult for me to figure out what's causing the error without a reproducible example on a dataset that I have access to.

[yt723]

There are 12091 cells.

Standard deviation of none of the genes was zero (minimum was 2.9e-6). I set up the genes manually, and it worked. I tested multiple std thresholds for selecting genes and eventually I selected genes with std >1e-2 (400/477 genes were selected).
I'm ok with the results. Thank you so much for your advice.

Yuko