In drewvoigt10/cellcuratoR: Shares Interactive Single-Cell Visualizations from Seurat

knitr::opts_chunk$set(
  collapse = TRUE,
  comment = "#>"
)

Overview

This vignette will walk through the exportation of a processed seurat object that is compatible with the cellcuratoR shiny visualization system. The code from this vignette does not compile on installation (as this vignette creates data directories and writes files to the user's computer). Therefore, it is recommended to copy and paste code chunks from this vignette to run in a local R session.

The export_seurat_object() function requires an S4 Seurat object. Compatible S4 Seurat input objects have either (A) been processed with Seuart (version 3) or (B) been updated to a version 3 Seurat object with the function UpdateSeuratObject(). Further, within the meta.data of the Seurat object, the user should have columns corresponding to a final cluster designation, a library-id, and a cell type classification, as explained below in this tutorial.

We will walk through the exportation of a stimulated vs control pbmc dataset freely available from Seurat. We will use the ifnb data, in which peripheral blood mononuclear cells were divided into stimulated and control groups. For further details, see https://satijalab.org/seurat/v3.1/immune_alignment.html. First, we will download the ifnb dataset using the helpful SeuratData package.

library(cellcuratoR)
library(Seurat)
library(tidyverse)

if(!requireNamespace("SeuratData", quietly = TRUE)){
  stop("Package \"SeuratData\" needed for this vignette to work. Please install it.",
      call. = FALSE)
}

library(SeuratData)

InstallData("ifnb") 
data("ifnb")

head([e-mail])

We will perform a basic integration and clustering analysis, as outlined by the Seurat team.

ifnb.list <- SplitObject(ifnb, split.by = "stim")

ifnb.list <- lapply(X = ifnb.list, FUN = function(x) {
    x <- NormalizeData(x)
    x <- FindVariableFeatures(x, selection.method = "vst", nfeatures = 2000)
})

immune.anchors <- FindIntegrationAnchors(object.list = ifnb.list, dims = 1:20)
immune.combined <- IntegrateData(anchorset = immune.anchors, dims = 1:20)
DefaultAssay(immune.combined) <- "integrated"

# Run the standard workflow for visualization and clustering
immune.combined <- ScaleData(immune.combined, verbose = FALSE)
immune.combined <- RunPCA(immune.combined, npcs = 30, verbose = FALSE)
# t-SNE and Clustering
immune.combined <- RunUMAP(immune.combined, reduction = "pca", dims = 1:20)
immune.combined <- FindNeighbors(immune.combined, reduction = "pca", dims = 1:20)
immune.combined <- FindClusters(immune.combined, resolution = 0.5)

# We map each cluster to a putative cell type
seurat_celltypes <- c("CD14 Mono", "CD4 Naive T", "CD4 Memory T", 
                      "CD16 Mono", "B", "CD8 T", "T activated", 
                      "NK", "DC", "B Activated", "Mk",  "pDC", "Eryth")
celltype <- plyr::mapvalues(x = [e-mail]$seurat_clusters, 
                                from = seq(from = 0, to = 12), 
                                to = seurat_celltypes)
[e-mail]$celltype <- celltype

We add an artificial treatment column to the seurat object so that differential expression can be performed between unstimulated and stimulated samples. Treatment must be a binary factor. In instances where there are multiple factor levels to a biological condition (eg early, mid, late disease stage), the user is encouraged to create seperate columns in the meta.data for each comparison of interest (eg early vs late, ealy vs mid, mid vs late), and specifying all of these column names in the additional_metadata_cols argument listed below.

treatment <- as.factor([e-mail]$orig.ident)
[e-mail] <- data.frame([e-mail], treatment)

Before interacting with the data, we must export the processed seurat object in a format interpretable by Shiny. First, we create a directory to organize all of our exported cellcuratoR objects:

## The user should consider changing filepath to ~/Desktop or other directory that is easier to navigate to
my_filepath <- Sys.getenv("R_LIBS_USER")

dir.create(file.path(my_filepath, "my_cellcuratoR_objects/"))

Next, we export our processed Seurat object into this newly created directory with the export_shiny_object() function.

export_shiny_object(seurat_object = immune.combined, 
                    final_cluster_column_name = "seurat_clusters",
                    library_id_column_name = "orig.ident",
                    classification_column_name = "celltype",
                    create_dendrogram = TRUE,
                    custom_cluster_colors = FALSE,
                    custom_library_colors = FALSE,
                    additional_metadata_cols = c("treatment"), # allows for dge between treatment groups
                    export_data_path = file.path(my_filepath, "my_cellcuratoR_objects/infb_shiny/"))
## If re-running locally, the user should consider changing filepath 
## to ~/Desktop or other directory that is easier to navigate to

The directory structure of the exported dataset is as follows:

|-- infb_shiny
|   |--seurat_obj.RData
|   |--seurat_obj_big.RData

Once exported, we can now launch the cellcuratoR app (cellcuratoR::launchApp()) and interact with the dataset. Upon launching the cellcuratoR shiny interface, we click the "Select Seurat Object Directory" button and navigate to our newly created directory, which by default in this vignette is the R library path, but can be changed if re-running locally with the export_data_path argument in export_shiny_object(). Importantly, we do NOT navigate to the directory of the individual dataset (eg, do NOT navigate to "~/Desktop/my_cellcurator_objects/infb_shiny/", DO navigate to "~/Desktop/my_cellcurator_objects/"). Then, we can load our dataset with the "which dataset should be loaded?" dropdown and initiate exploratory data analysis.

print(paste0("exporting data to: ", my_filepath, "/my_cellcuratoR_objects/"))
# Launch the app with the following command (commented out for R-markdown)
# cellcuratoR::launchApp()

# 1. Navigate to the our newly created directory printed above (or change to a more accesible directory for you)
# 2. Select the "infb_shiny" dataset from the "which dataset should be loaded?" dropdown
# 3. Interactively explore data!

drewvoigt10/cellcuratoR documentation built on May 22, 2023, 4:38 p.m.