sipsic_find_clustering

1 Functions

library(GSEABase)
filter_pathways <- function(pathway,dataset) {
  genes_in_data = pathway@geneIds %in% rownames(dataset) %>% sum()
  pathway@setName = paste( pathway@setName,genes_in_data,sep = "_genes:") #if genes in data are less than threshold, mark them
  pathway
}

calculatePathwayScores <- function(pathwayToScore, countsMatrix)
{
  pathwayName <- pathwayToScore@setName
  print(pathwayName)
  pathwayGenes <- pathwayToScore@geneIds
  pathwayScoresObject <- getPathwayScores(countsMatrix, pathwayGenes) 
  suppressWarnings(
    if(is.na(pathwayScoresObject)){return (NA)}
  )
  return(pathwayScoresObject$pathwayScores)
}

sipsic_pipeline <- function(geneset, dataset,thresholds) {

  pathwayScoreLists <- lapply(X = geneset@.Data, calculatePathwayScores, dataset %>% as.matrix() %>% Matrix(sparse = T))
  pathwayScoreLists <- pathwayScoreLists[!is.na(pathwayScoreLists)] #remove NA pathways
  pathwayScoresMatrix <- as.data.frame(do.call("rbind", pathwayScoreLists))
  for(currPathwayIndex in 1:length(pathwayScoreLists)){
    rownames(pathwayScoresMatrix)[currPathwayIndex] <- geneset@.Data[[currPathwayIndex]]@setName
  }
  
  seurat_assay <- CreateAssayObject(counts = pathwayScoresMatrix)
  return(seurat_assay)
}

2 Data

GBMSeurat_cancer = readRDS("./Data/GSM3828673_10X_GBM_seurat_cancer.RDS")
genesets_h <- getGmt("./Data/msigdb_pathways/h.all.v7.0.symbols.pluscc.gmt")
genesets_cp <- getGmt("./Data/msigdb_pathways/c2.cp.v2023.2.Hs.symbols.gmt")
genesets_kegg <- getGmt("./Data/msigdb_pathways/c2.cp.kegg_medicus.v2023.2.Hs.symbols.gmt")
genesets_pid<- getGmt("./Data/msigdb_pathways/c2.cp.pid.v2023.2.Hs.symbols.gmt")

genesets_test = GeneSetCollection(genesets_h[10:13])
genesets_h_cp = GeneSetCollection(c(genesets_h,genesets_cp))
genesets_h_kegg = GeneSetCollection(c(genesets_h,genesets_kegg))
genesets_h_pid = GeneSetCollection(c(genesets_h,genesets_pid))

GBMSeurat_cancer@assays$RNA <- data.frame()
saveRDS(GBMSeurat_cancer,file = "./Data/GSM3828673_10X_GBM_seurat_cancer_msigdb_pathwyas.RDS")

3 original Louvain algorithm

print_tab(all_p_patient,title = "patients barplot")

patients barplot

print_tab(all_p_cellType,title = "cancer type barplot")

cancer type barplot

print_tab(all_p_umap,title = "UMAPs")

UMAPs

NA

4 Louvain algorithm with multilevel refinement

all_p

5 SLM algorithm

all_p

6 Leiden algorithm

all_p