scPipeline

• Pipeline for basic end-to-end scRNA-seq processing

Installation:

devtools::install_github("jackbibby1/scPipeline")

Main functions:

• read_scrna() covers:
  • Reading in data, either cellranger, h5, or tab files (multicore option)
  • Filtering based on mito percentages
  • Prompt user for input on mito cutoff before proceeding
  • Adding metadata
  • Merging samples if >1
• norm_integration() covers:
  • Normalisation (Log or SCT)
  • Getting variable features
  • Scaling
  • Dimensionality reduction (PCA)
  • Batch correction and integration (any method available in Seurat::IntegrateLayers())
• seurat_clustering()
  • Defining number of PCs
  • Dimensionality reduction (UMAP and/or tSNE)
  • Clustering

Example:

# set wdir and generate metadata ------------------------------------------
setwd("path-to/wdir")
folders <- list.dirs("data", recursive = F)
meta <- data.frame(time = rep(c(0, 12, 24), times = 4),
                   cell = rep(c("cd4", "cd8"), each = 6),
                   lineage = rep(c("naive", "memory"), each = 3, times = 2))

# reading in sc data ---------------------------------------------------
# generates single Seurat object containing data merged from all filepath folders, with annotated metadata
df <- read_scrna(filepath = "~/Desktop/test/data",
                 filename_pattern = "GSM",
                 cores = 4, 
                 metadata = meta,
                 mito_pattern = "^MT-",
                 merge_data = TRUE)
                 
# normalisation and batch correction ---------------------------------------------------           
# performs normalisation, var features finding, scaling, PCA and joins layers
df <- norm_integration(df,
                       normalisation_method = "LogNormalize",
                       batch_correction = T,
                       correction_method = "HarmonyIntegration")
                 

# clustering data ---------------------------------------------------
# generates single Seurat object with umap and clustering performed

df <- seurat_clustering(df, reduction = "integrated")