Error at computeCommunProbPathway

[tkapello]

Hi,

thank you for this nice package. I am encountering an issue which has been archived in the old github repository. I expanded the CellChat database with the Omnipath resource and I ran CellChat on a seurat object. When I run computeCommunProbPathway(cellchat), I get the following error:

Error in prob.pathways[, , pathways.sig] : subscript out of bounds

I attach below the code I have run and I am happy to share the object if needed.

cellchat<-createCellChat(object=seurat@assays$RNA@data, [email protected], group.by='annotation')

# Load Cellchat database
cellchatdb<-CellChatDB.human

# Assemble Omnipath databases
intercell<-import_intercell_network(organism=9606)
intercell<-intercell[intercell$entity_type_intercell_target=='protein' & intercell$entity_type_intercell_source=='protein',]
ptm<-import_omnipath_enzsub(organism=9606)

intercell<-intercell[,c(8,7)]
colnames(intercell)<-c('ligand', 'receptor')
ptm<-ptm[,c(3,4)]
colnames(ptm)<-c('ligand', 'receptor')

interactions<-rbind(intercell, ptm)
interactions$interaction_name<-paste(interactions$ligand, sep='_', interactions$receptor)
interactions<-interactions[!duplicated(interactions$interaction_name),]
interactions<-as.data.frame(interactions)
rownames(interactions)<-interactions$interaction_name
interactions$interaction_name_2<-gsub('_', ' - ', interactions$interaction_name)
interactions[, c("agonist", "antagonist", "co_A_receptor", "co_I_receptor", "evidence", "annotation", "pathway_name")] <- NA
interactions<-interactions[,c(3, 11, 1, 2, 5:10, 4)]
interactions[]<-lapply(interactions, function(x) ifelse(is.na(x), "", x))
interactions<-rbind(interactions, cellchatdb$interaction)
interactions<-interactions[!duplicated(interactions$interaction_name),]

complex<-import_intercell_network(organism=9606)
complex<-complex[complex$entity_type_intercell_source=='complex' | complex$entity_type_intercell_target=='complex',]
complex<-complex %>% distinct(source_genesymbol, .keep_all=TRUE)
complex<-separate(complex, source_genesymbol, into = c("subunit_1", "subunit_2", "subunit_3", "subunit_4"), sep = "_")
complex<-as.data.frame(complex)

for(i in complex$genesymbol_intercell_source){
  if(grepl(':', i)==TRUE) {
    complex[complex$genesymbol_intercell_source==i, "genesymbol_intercell_source"]<-sapply(strsplit(split=':', i), '[', 2)
  } else {
    complex[complex$genesymbol_intercell_source==i, "genesymbol_intercell_source"]<-i
    }
}
  
rownames(complex)<-complex$genesymbol_intercell_source
complex<-complex[,8:11]
complex[]<-lapply(complex, function(x) ifelse(is.na(x), "", x))
complex<-rbind(complex, cellchatdb$complex)

gene_symbols<-unique(c(interactions$ligand, interactions$receptor, 
                       complex$subunit_1, complex$subunit_2, complex$subunit_3, complex$subunit_4,
                       cellchatdb$cofactor$cofactor1, cellchatdb$cofactor$cofactor2, cellchatdb$cofactor$cofactor3, cellchatdb$cofactor$cofactor4, cellchatdb$cofactor$cofactor5, 
                       cellchatdb$cofactor$cofactor6, cellchatdb$cofactor$cofactor7, cellchatdb$cofactor$cofactor8, cellchatdb$cofactor$cofactor9, cellchatdb$cofactor$cofactor10,
                       cellchatdb$cofactor$cofactor11, cellchatdb$cofactor$cofactor12, cellchatdb$cofactor$cofactor13, cellchatdb$cofactor$cofactor14, cellchatdb$cofactor$cofactor15,
                       cellchatdb$cofactor$cofactor16))
gene_symbols<-gene_symbols[!is.na(gene_symbols)]

gene_info<-select(org.Hs.eg.db, keys=gene_symbols, keytype="SYMBOL", column=c("GENENAME", "ENTREZID", "ENSEMBL"))
colnames(gene_info)<-c('Symbol', 'Name', 'EntrezGene.ID', 'Ensembl.Gene.ID')

omnipath<-list(interactions=interactions, complex=complex, cofactor=cellchatdb$cofactor, geneInfo=gene_info)

# Add Omnipath database to CellChat object
cellchat@DB<-omnipath

# Pre-process datasets
cellchat<-subsetData(cellchat)
cellchat<-identifyOverExpressedGenes(cellchat)
cellchat<-identifyOverExpressedInteractions(cellchat)

# Compute cell-cell communication probability
cellchat<-computeCommunProb(cellchat, population.size=T)
cellchat<-filterCommunication(cellchat, min.cells=50)

# Compute signaling pathway probability
cellchat<-computeCommunProbPathway(cellchat)

I would appreciate any help!

[sqjin]

@tkapello This issue is clear. You have to provide the pathway information for each L-R pair when running computeCommunProbPathway

[tkapello]

Thank you @sqjin for the reply. May I ask how you annotated the pathways when you were building the tool? It seems to me that you name each pathway by the ligand. Did you have any rules?