In vondoRishi/scpackages: scpackages has many single-cell R and python packages to use them easily.

library(scpackages)

Process Seurat Object for clustering

First we will load saved seurat object from disk.

seuratObj <- readRDS("/home/dasroy/data/Project/scpackages/1k_Brain_E18_Mm.12_February_2020.rds")

3-in-1

There are three steps need to do at first ; which are

• Normalization
• VariableFeatures (find variable genes)
• ScaleData

These steps can be done by a single method with default values. It will also genarate VariableFeaturePlot. Only the variable features will be scaled as those are only required as input for PCA.

Note We can pass any parameter of above "Seurat" functions to below function. Here in this example nfeatures belongs to VariableFeatures .

seuratObj <- normScaleHVG(seuratObj,seuratVerbose = FALSE,nfeatures=1000)

To save the variance info of all the genes in a csv file.

tmp_hvf <-HVFInfo(seuratObj)
tmp_hvf$Symbol <- rownames(tmp_hvf)
write.csv(tmp_hvf, file = "HVF_info.csv",row.names = FALSE)

CellCycle

Next we will check

library(readr)
cell_cycle_genes <- read_csv(system.file("extdata", "cell_cycle_genes.csv", package = "scpackages"))
cell_cycle_genes

NOTE cell_cycle_genes should have columns "Symbol" for gene names and "Phase" with values (S and M) for cell-cycle phases

Regression

At this step we will check

• Principal components(PC) are loaded with cell-cycle (CC) genes.
• PCA plot with CC genes
• regressing out CC effect with Seurat method CellCycleScoring
• Again PCA plot with CC genes

seuratObj <-qc_regress_CellCycle(seuratObj, cell_cycle_genes)

Plots will be saved in this following slots.

• 'seuratObj@misc$CC_pca_bar'

• 'seuratObj@misc$before_cc_pca'

• 'seuratObj@misc$after_cc_pca'

Cluster

Elbowplot

seuratObj <- pcaProcess(seuratObj,features = VariableFeatures(seuratObj),jackStraw = FALSE)
# seuratObj@misc$elbowPlot

Clustering

seuratObj <- makeClusterSeurat(seuratObj,maxDims = 20, res = 0.5)
# seuratObj@misc$umapPlot

Cluster markers

markers_info <-  FindAllMarkers( object = seuratObj,
                                 only.pos = TRUE, min.pct = 0.25,
                                 thresh.use = 0.25, verbose = FALSE )
head(markers_info)
seuratObj@misc$markers_info <- markers_info

To save the markers_info in a csv file.

write.csv(markers_info, file = "cluster_markers.csv",row.names = FALSE)

Cell type information

Dummy cell type info. This data source must have two columns "Symbol" and "Gene_type"

marker_gene <- read_csv("/home/dasroy/data/Project/GeneExpression/scrna_workflow/marker_genes.csv")
marker_gene <- marker_gene[!duplicated(marker_gene$Symbol),]
head(marker_gene)

Annotation

library(dplyr)
markers_info <- left_join(markers_info,marker_gene, 
                          by = c("gene" = "Symbol")) %>% 
    replace_na(list(Gene_type = "Non skin"))

# Storing the info 
seuratObj@misc$markers_info <- markers_info

Rename clusters

seuratObj <- renameCluster(ClusterInfo = markers_info, Object = seuratObj)

We used genes which should express in developing skin and hence all the clusters are marked with non-skin genes. The third field of the cluster names reflect the cluster sizes.

Report

It will genarate report and save seuratObj with filenames begining with title and ending with date_stamp.

report_Cluster(obj_scRNA = seuratObj,title = "../scaled_cluster")

vondoRishi/scpackages documentation built on Feb. 27, 2020, 1:52 a.m.