scutilsR

The package scutilsR is my code hub for utilities in scRNA-seq analysis.

Installation

You can install the development version of scutilsR from GitHub with:

# install.packages("devtools")
# install DoubletFinder
devtools::install_github("chris-mcginnis-ucsf/DoubletFinder", ref = "554097b")
# install CellChat
devtools::install_github("sqjin/CellChat", ref = "418b660")
devtools::install_github("JarningGau/scutilsR")

Example

IO

Read output from STARsolo

Elife_2019_mouse_GSE113293/ # sample hub: contains a list of samples
├── GSM3102982              # sample ID:  contains a list of assays
│   ├── corrected
│   │   ├── barcodes.tsv
│   │   ├── decontX.info.tsv
│   │   ├── features.tsv
│   │   └── matrix.mtx.gz
│   ├── filtered            # output of STARsolo after cell calling
│   │   ├── barcodes.tsv
│   │   ├── features.tsv
│   │   └── matrix.mtx.gz
│   └── raw                 # output of STARsolo before cell calling
│       ├── barcodes.tsv
│       ├── features.tsv
│       └── matrix.mtx.gz

mm <- ReadSolo(path = "/path/to/samplehub", assay = "assay")

Read output from DNBelab C4 pipeline

data <- ReadC4("/path/to/dnbc4.txt.gz") # C4 output is a compressed dense matrix

Doublets Removal

Mark doublets rather than remove them via DoubletFinder.

MarkDoublets() function run DoubletsFinder separately. All parameters for DoubletsFinder are default.

• pK: auto selected by FindOptimalpK()
• pN: 0.25
• estimated percentage of doublets: 0.075

seu <- MarkDoublets(seu = seu, PCs = 1:10, split.by = "orig.ident")

Users can remove the marked doublets or clusters enriched marked doublets.

Remove Ambient RNAs & Calculate Contimination Rates

If you have the clusters information.

seu <- RemoveAmbientRNAs(seu, split.by = "orig.ident", cluster.name = "seurat_clusters")

else

seu <- RemoveAmbientRNAs(seu, split.by = "orig.ident", cluster.name = NULL)

Find All Markers Parallelly

all.markers <- mcFindAllMarkers(seu = seu, do.flatten = T, only.pos = T, n.cores = 10) # returns a data.frame
all.markers <- mcFindAllMarkers(seu = seu, do.flatten = F, only.pos = T, n.cores = 10) # returns a list

Annotate Cell Types According to Markers via Enrichment Analysis

collected gene sets

• Mouse Cell Atlas (MCA)
• Human Cell Lanscape (HCL)

all.markers <- mcFindAllMarkers(seu.ds, do.flatten = F, only.pos = T, n.cores = 20)
all.markers <- lapply(all.markers, function(xx) subset(xx, p_val_adj < 1e-10 & avg_log2FC > log2(2))$Gene.name.uniq)

## enrichment analysis (human)
data("mca_hsa")
data("hcl_hsa")
t2g <- rbind(mca_hsa, hcl_hsa)

e.res <- enrich_batch(all.markers, t2g)
enrich_dotplot(e.res)

Data Imputation

DefaultAssay(seu) <- "RNA"
seu <- impute_nmf(seu, min_cells = 50, k = 100, threads = 10, seed = 1024)

Cell-Cell Communication

A CellChat wrapper CellChatHelper() was implemented in scutilsR

DB <- subsetDB(CellChat::CellChatDB.human, search = "Secreted Signaling")

# outputs:
# - {out.dir}/cellchat.{name}.{mode}.rds
# - {out.dir}/cellchat.{name}.{mode}.pdf

CellChatHelper(seu = seu, 
               label.field = "seurat_clusters", 
               name = "run_name", 
               mode = "default", 
               DB = DB, 
               out.dir = getwd(), 
               cores = 10, 
               fig.width = 10, 
               fig.height = 8)