Install RCSL package and other requirements

RCSL can be installed directly from GitHub with ‘devtools’.

library(devtools)
devtools::install_github("QinglinMei/RCSL")

Now we can load RCSL. We also load the SingleCellExperiment, ggplot2 and igraph package.

library(RCSL)
library(SingleCellExperiment)
library(ggplot2)
library(igraph)
library(umap)

Load dataset (yan)

We illustrate the usage of RCSL on a human preimplantation embryos and embryonic stem cells(Yan et al., (2013)). The yan data is distributed together with the RCSL package, with 90 cells and 20,214 genes:

data(yan, package = "RCSL")
head(ann)

##                 cell_type1
## Oocyte..1.RPKM.     zygote
## Oocyte..2.RPKM.     zygote
## Oocyte..3.RPKM.     zygote
## Zygote..1.RPKM.     zygote
## Zygote..2.RPKM.     zygote
## Zygote..3.RPKM.     zygote

yan[1:3, 1:3]

##          Oocyte..1.RPKM. Oocyte..2.RPKM. Oocyte..3.RPKM.
## C9orf152             0.0             0.0             0.0
## RPS11             1219.9          1021.1           931.6
## ELMO2                7.0            12.2             9.3

origData <- yan
label <- ann$cell_type1

1. Pre-processing

In practice, we find it always beneficial to pre-process single-cell RNA-seq datasets, including: 1. Log transformation. 2. Gene filter

data <- log2(as.matrix(origData) + 1)
gfData <- GenesFilter(data)

2. Calculate the initial similarity matrix S

resSimS <- SimS(gfData)

## Calculate the Spearman correlation 
## Calculate the Nerighbor Representation 
## Find neighbors by KNN(Euclidean)

3. Estimate the number of clusters C

Estimated_C <- EstClusters(resSimS$drData,resSimS$S)

## ======== Calculate maximal strongly connected components ======== 
## ======== Calculate maximal strongly connected components ======== 
## ======== Calculate maximal strongly connected components ========

4. Calculate the block diagonal matrix B

resBDSM <- BDSM(resSimS$S, Estimated_C)

## ======== Calculate maximal strongly connected components ========