clustView

A shiny app for browsing and annotating scRNAseq clusters.

This package is still experimental!

Next on the to-do list:

• generalization to SingleCellExperiment objects
• functionality to merge clusters
• importing cell type predictions based on references

Installation

Required R packages for the app:

BiocManager::install(c( "Seurat", "shiny","shinydashboard","shinycssloaders",
   "DT", "data.table", "ggplot2", "cowplot", "AnnotationDbi", "GO.db" ) )

Package clustree is also required:

devtools::install_github("lazappi/clustree", dependencies = TRUE)

In addition, preparing the cluster annotation will require the appropriate org.Xx.eg.db for your species, e.g. org.Hs.eg.db.

I tried to make it backward-compatible with Seurat v2, but right now it's only tested with V3. To install Seurat V3:

devtools::install_github(repo = 'satijalab/seurat', ref = 'release/3.0')

To install ClustView:

devtools::install_github("plger/clustView")

Usage

Preparing cluster annotation:

Assuming a seurat object named se, first get all markers for each cluster/resolution. For example, in v3 (assuming the default prefix after integration):

library(clustView)

# get all computed resolutions:
cn <- names(seurat@meta.data)
resolutions <- as.numeric(gsub("integrated_snn_res.","",cn[grep("^integrated_snn_res",cn)],fixed=T))

# get markers for each cluster/resolution:
markers <- list()
for(r in resolutions){
  markers[[as.character(r)]] <- FindAllMarkers( 
	object = se,
	only.pos = FALSE, 
	min.pct = 0.25, 
	resolution = r,
	logfc.threshold = 0.5,
	test.use = "wilcox",
	max.cells.per.ident = 300 )
}

se <- prepSeuratForClustView( se,
                              markerslist=markers,
                              species="Hs",
                              ontologies=c("BP", "CC") )

Launching the app:

Assuming a seurat object named se:

clustView(se)