In rikenbit/scTensor: Detection of cell-cell interaction from single-cell RNA-seq dataset by tensor decomposition

Summary of the output objects of scTensor

Here, we introduced the objects saved in reanalysis.RData.

library("scTensor")
load("reanalysis.RData")

After performing cellCellReport, some R objects are saved in the reanalysis.RData as follows;

• sce : SingleCellExperiment object
  • metadata(sce)$lrbase : The file pass to the database file of LRBase
  • metadata(sce)$color : The color vector specified by cellCellSetting
  • metadata(sce)$label : The label vector specified by cellCellSetting
  • metadata(sce)$algorithm : The algorithm for performing r Biocpkg("scTensor")
  • metadata(sce)$sctensor : The results of r Biocpkg("scTensor")
    • metadata(sce)\$sctensor\$ligand : The factor matrix (Ligand)
    • metadata(sce)\$sctensor\$receptor : The factor matrix (Receptor)
    • metadata(sce)\$sctensor\$lrpair : The core tensor
  • metadata(sce)$datasize : The data size of CCI tensor
  • metadata(sce)$ranks : The number of lower dimension in each direction of CCI tensor
  • metadata(sce)$recerror : Reconstruction Error of NTD
  • metadata(sce)$relchange : Relative Change of NTD
• input : The gene expression matrix <# Genes * # Cells>
• twoD : The result of 2D dimensional reduction (e.g. t-SNE)
• LR : The Ligand-Receptor corresponding table extracted from LRBase.XXX.eg.db
• celltypes : The celltype label and color scheme
• index : The core tensor values
• corevalue : The core tensor values (normalized)
• selected : The selected corevalue position with thr threshold "thr"
• ClusterL : The result of analysis in each L vector
• ClusterR : The result of analysis in each R vector
• out.vecLR : The result of analysis in LR pairs
• g : The igraph object to visualize ligand-receptor gene network

Execution of scTensor with the different options

Using the reanalysis.RData, some users may want to perform r Biocpkg("scTensor") with different parameters.

For example, some users want to perform cellCellDecomp with different ranks, perform cellCellReport with omitting some enrichment analysis, provide the results to their collaborators.

To do such tasks, just type like belows.

library("AnnotationHub")
library("LRBaseDbi")

# Create LRBase object
ah <- AnnotationHub()
dbfile <- query(ah, c("LRBaseDb", "Homo sapiens", "v002"))[[1]]
LRBase.Hsa.eg.db <- LRBaseDbi::LRBaseDb(dbfile)

# Register the file pass of user's LRBase
metadata(sce)$lrbase <- dbfile(LRBase.Hsa.eg.db)

# CCI Tensor Decomposition
cellCellDecomp(sce, ranks=c(6,5), assayNames="normcounts")

# HTML Report
cellCellReport(sce, reducedDimNames="TSNE", assayNames="normcounts",
    title="Cell-cell interaction within Germline_Male, GSE86146",
    author="Koki Tsuyuzaki", html.open=TRUE,
    goenrich=TRUE, meshenrich=FALSE, reactomeenrich=FALSE,
    doenrich=FALSE, ncgenrich=FALSE, dgnenrich=FALSE)

Session information {.unnumbered}

sessionInfo()

rikenbit/scTensor documentation built on Jan. 28, 2023, noon