scrnaseq_pipeline

This is a complete pipeline for analysing scRNA-seq datasets from 10x Genomics.

The workflow uses Seurat (v3.0.2) and performs all necessary steps for an basic exploratory analysis of your data. These include:

• normalization and scaling of gene-expression matrix,
• filtering of low quality cells and outliers,
• dimensionality reduction (PCA, TSNE, and UMAP),
• clustering (using different resolutions),
• and differential expression analysis.

Seurat offers a wide variety of functions and tutorials that allow you to perform exhaustive exploration of your scRNA-seq dataset. For this reason, the present pipeline is essentially following the same workflow provided by Satija lab (Seurat - Getting Started)

The advantage of this pipeline is that it allows for automatised exploratory analysis of your data, using a one-liner on the CLI.

Rscript script.R -i path/to/input -o path/to/output 

In order to use this pipeline you will need to download the repo to your local and set your working directory.

• Click on Clone or Download
• Download ZIP
• Change working directory

cd /path/to/repo 

Overview

The pipeline is composed of two executable R scripts:

• explore.R Performs all the exploratory analysis (from loading the gene-expression matrix to UMAP and clustering).
• cluster.R Performs differential expression analysis based on the selected resolution used for clustering.

Run Rscript explore.R -h to see arguments to be passed with the explore.R script:

Usage: explore.R [options]


Options:
	-i INPUT, --input=INPUT
		Set the path to the directory containing
              matrix.mtx, genes.tsv, and barcodes.tsv files. 
              Make sure files are unzipped

	-o OUTPUT, --output=OUTPUT
		Specify output directory for storing plots and tables

	-g GENES, --genes=GENES
		(Optional) Specify path to a .csv file listing genes

	-c CELLCYCLE, --cellcycle=CELLCYCLE
		(Optional) If TRUE, calculates cell cycle score using 
              S- and G2M-related genes

	-h, --help
		Show this help message and exit

Specify -o path/to/out to create an out directory where all plots will be saved. When -g genes.csv is added, the script returns a .pdf file displaying gene expression levels on a UMAP embedding. Gene lists should be saved in the repo directory, as a one-column .csv file. E.g.:

gene
ASCL1
DCX
MKI67

Run Rscript cluster.R -h to see arguments to be passed with the cluster.R script:

Usage: cluster.R [options]


Options:
	-i INPUT, --input=INPUT
		Set the path to the R objects previously saved.

	-r RESOLUTION, --resolution=RESOLUTION
		Set the resolution for the clustering algorithm (i.e., from 0.1 to 1

	-o OUTPUT, --output=OUTPUT
		(Optional) Specify output directory for storing plots and tables.

	-g GENES, --genes=GENES
		(Optional) Specify path to a .csv file listing genes 

	-h, --help
		Show this help message and exit

Use -oto save output files from cluster.R in a new directory. If not specified, files will be saved in the out directory created while running explore.R.

Prerequisites

• Install the latest R (I have used R v3.6.0)
• Packages needed for running the pipeline will automatically be installed, if not already present. If Seurat is already installed, make sure to upgrate to version >= 3.0.0.
• Input option requires the gene-expression files from CellRanger to be unzipped (e.g., gunzip path/to/files/*.gz and named as: barcodes.tsv, genes.tsv, matrix.mtx.

Output

Output files for explore.R include:

• Exploratory plots:
  • PrePostFilterVln.pdf
  • HVG.pdf
  • tSNEPlot.pdf
  • UMAPPlot.pdf
  • CellCycleScore.pdf
• Exploratory tables:
  • PCACellEmbeddings.csv
  • PCAFeatureLoadings.csv
  • tSNECellEmbeddings.csv
  • UMAPCellEmbeddings.csv
• Seurat objects:
  • prefilter.rds
  • filtered.rds
  • scaled.rds
  • tsne.rds
• (Optional) Feature plots (FeaturePlot.pdf).

Output files for cluster.R include:

• (Optional) Feature plots (FeaturePlot.pdf)
• List of cluster markers (ClusterMarkers.csv)
• Top 20 cluster markers (Top20ClusterMarkers.csv)
• Heatmap with the top 20 features per cluster (Top20MarkerHM.pdf)

Progress

Use > progress.log 2>&1 to build a progress report. E.g.,

Rscript cluster.R -i path/to/input -g genes.csv > progress.log 2>&1