Pre-processing

2023-06-11

We recommend CellFunTopic package to perform data pre-processing for convenience. The CellFunTopic package provides a convenient workflow for data pre-processing (quality control, normalization, dimension reduction, clustering, differential expression analysis, etc.) by integrating methods of Seurat package.

devtools::install_github("compbioNJU/CellFunTopic")
library(CellFunTopic)

Load in the data

CellFunTopic allows various types of input, including CellRanger outputs, raw gene expression matrix, and popular R objects used for single cell analysis such as SingleCellExperiment, Seurat, CellDataSet. CellFunTopic will transform different input data into a Seurat object.

Let us use a randomly generated tiny single-cell expression matrix of Arabidopsis thaliana as a toy example.

library(scPlant)
dim(example_Ath) # an internal example data of scPlant 

## [1] 7000 2000

SeuratObj <- readData(data = example_Ath, 
                      type = 'expMatrix', 
                      species = "Arabidopsis thaliana")

You can also take CellRanger outputs as input, for example:

SeuratObj <- readData(data = "filtered_gene_bc_matrices/TAIR/", 
                      type = '10X', 
                      species = "Arabidopsis thaliana")

Standard pre-processing workflow

Then we can perform pre-processing conveniently. If users provide a pre-processed Seurat object, this step can be skipped.

SeuratObj <- QCfun(SeuratObj)
SeuratObj <- RunSeurat(SeuratObj, resolution = 0.3)

unique(Seurat::Idents(SeuratObj)) # see how many clusters we got.

## [1] 1 0 3 2 5 4 6 7
## Levels: 0 1 2 3 4 5 6 7

Seurat::DimPlot(SeuratObj, reduction = "umap")