computeCommunProb: Error in if (sum(P1) == 0) {: missing value where TRUE/FALSE needed

[radiasso]

Hi! I'm new to CellChat (just installed recently), but I can't seem to make it work. A bit of context: I have an h5ad data object from 10x scRNA experiments (batch correction for multiple samples). My anndata object has cells as columns, genes as rows, and a normalized and log1p expression matrix in X. I also have the raw counts in adata.obs['n_counts'].

I followed the tutorial here starting from AnnData objects, and here for the following parts.

My lines of code specifically are:

library(anndata)
ad <- read_h5ad("ad.h5ad")
data.input <- t(as.matrix(ad$X)).  #since I already have the norm log1p data in X

meta <- ad$obs
meta$labels <- meta$Cell_types.  #I know probably redundant

library(CellChat)
cellchat <- createCellChat(object = data.input, meta = meta, group.by = "labels")
CellChatDB <- CellChatDB.mouse
showDatabaseCategory(CellChatDB)
CellChatDB.use <- CellChatDB
cellchat@DB <- CellChatDB.use
cellchat <- subsetData(cellchat)
cellchat <- identifyOverExpressedGenes(cellchat)
cellchat <- identifyOverExpressedInteractions(cellchat)

The problem arises right after, when launching: cellchat <- computeCommunProb(cellchat, type = "triMean") At first it seems to work but then it outputs: Error in if (sum(P1) == 0) {: missing value where TRUE/FALSE needed

I can't figure out what bothers that function. I've looked through issues and questions here on Git but I still have no clue. Can anyone help me please? Thank you in advance, so so much!

[sqjin]

@radiasso Can you check the updated tutorial

[radiasso]

@radiasso Can you check the updated tutorial

Hi! Thank you for your answer. Unfortunately, I had followed that tutorial and I tried again right now, but the error is always the same when I try to launch computeCommunProb.

> library(anndata)
> ad <- read_h5ad("ad.h5ad")
> data.input <- t(as.matrix(ad$X))  #my normalized and log1p data is in X already
> meta <- ad$obs 
> meta$labels <- meta[["Cell_types"]] 
> library(CellChat)
> cellchat <- createCellChat(object = data.input, meta = meta, group.by = "labels")
[1] "Create a CellChat object from a data matrix"
Set cell identities for the new CellChat object 
The cell groups used for CellChat analysis are  Collecting Duct, DCT-CNT, Endothelial, Leucocytes, Loop of Henle, PEC, PT-S1, PT-S2, PT-S3 

> CellChatDB <- CellChatDB.mouse
> showDatabaseCategory(CellChatDB)
> CellChatDB.use <- CellChatDB
> cellchat@DB <- CellChatDB.use
> cellchat <- subsetData(cellchat)
> cellchat <- identifyOverExpressedGenes(cellchat)
> cellchat <- identifyOverExpressedInteractions(cellchat)
  |++++++++++++++++++++++++++++++++++++++++++++++++++| 100% elapsed=00s  
  |++++++++++++++++++++++++++++++++++++++++++++++++++| 100% elapsed=00s  
  |++++++++++++++++++++++++++++++++++++++++++++++++++| 100% elapsed=01s  
The number of highly variable ligand-receptor pairs used for signaling inference is 1027 

> cellchat <- computeCommunProb(cellchat, type = "triMean")
triMean is used for calculating the average gene expression per cell group. 
[1] ">>> Run CellChat on sc/snRNA-seq data <<< [2024-02-19 14:12:48]"
  |                                                                      |   0%Error in if (sum(P1_Pspatial) == 0) { : 
  missing value where TRUE/FALSE needed

Basically the only different thing from the tutorial you linked is > meta$labels <- meta[["Cell_types"]], but the rest is the same. If it helps, I get warnings(), it says In log(x) : NaNs produced.

Same thing happens when I start from the sce object with zellkonverter. How can I fix this please?

[sqjin]

@radiasso Can you show data.input[1:10, 1:10]?