seurat-scripts 4.x.y for Seurat 4.4.0

In order to wrap Seurat's internal workflow in any given workflow language, it's important to have scripts to call each of those steps, which is what this package provides.

This version of seurat-scripts uses native conversions to Loom, SCE, Seurat H5 and AnnData.

Please note that given the little man power available to maintain this, we dropped MacOS support for Seurat version 4.4.0 onwards (not all tests were passing and there was no bandwidth to take care of the dependencies issues on both linux and macOS).

Install

The recommended method for usage is through the biocontainer produced by Bioconda. If needed, scripts installation is via a Bioconda recipe called seurat-scripts.

With the Bioconda channels configured the latest release version of the package can be installed via the regular conda install command:

conda install seurat-scripts

Test installation

There is a test script included:

r-seurat-scripts-post-install-tests.sh

This downloads a number of datasets and executes all of the scripts described below and more.

Commands

Currently wrapped Seurat functions are described below. Each script has usage instructions available via --help, consult function documentation in Seurat for further details.

These instructions might be currently outdated for Seurat 4.0.0, the best source of information is --help on each command.

Read10X(): read 10X data and create Seurat object from the matrix

seurat-read-10x.R -d <10x data directory> -o <output matrix object in .rds format>

FilterCells(): filter out poor-quality cells

seurat-filter-cells.R -i <raw Seurat object in .rds format> -s nGene,nUMI -l <min_genes>,<min_umi> -o <output Seurat object in .rds format>

NormalizeData(): normalise the expression values

seurat-normalise-data.R -i <filtered Seurat object in .rds format> -a <assay type> -n <normalisation method> -s <scale_factor> -o <normalised Seurat object in .rds format>

FindVariableGenes(): find variable genes

seurat-find-variable-genes.R -i <normalised Seurat object in .rds format> -m <mean function> -d <dispersion function> -l <fvg x low cutoff> -j <fvg x high cutoff> -y <fvg y low cutoff> -z <fvg y high cutoff> -o <output Seurat object in .rds format> -t <variable genes list in text format>

ScaleData(): scale expression values

seurat-scale-data.R -i <Seurat object with variable genes, in .rds format> -e <test genes> -v <variables to regress> -m <model to use> -u <use umi> -s <do scale> -c <do center> -x <scale max> -b <block size> -d <min cells to block> -a <assay type> -n <check for norm> -o <output Seurat object in .rds format>

RunPCA(): run a principal components analysis

seurat-run-pca.R -i <Seurat object with scaled expression valus in .rds format> -e <test genes> -p <pcs to compute> -m <use imputed> -o <output Seurat object in .rds format> -b <pca embedding in text format> -l <pca loadings file in format> -s <pca stdev file, text format>

FindTransferAnchors(): project a reference on a query (integration)

seurat-find-transfer-anchors.R -i <seurat object with computed dimension reduction used as query, .rds format> -r <seurat object with computed dimension reduction used as reference, .rds format> -o <anchorSet object with anchor matrix, .rds format> -n <normalization method: pcaproject or cca> -f <features to use for dimensional reduction> -d <which dimensions to use from the reduction to specify the neighbor search space, a:b format>    

DimPlot(): plot dimension reductions

seurat-dim-plot.r -i <Seurat object with computed dimension reductions, .rds format> -r <dimension reduction, e.g. pca> -a <dim 1> -b <dim 2> -p <pt size> -l <label size> -d <do label> -f <group by> -t <title> -w <png width> -j <png height> -o <image file>

RunTSNE(): run-tSNE analysis

seurat-run-tsne.r -i <Seurat object with computed PcA, .rds format> -r <reduction type> -d <dims to use> -e <file with genes to use> -f <do fast tsne> -o <output Seurat object in .rds format> -b <tsne embeddings in csv format>

FindClusters(): generate cell clusters from expression values

seurat-find-clusters.r -i <<Seurat object with computed dimension reductions, .rds format>> -e <test genes> -u <dimension reduction, e.g. pca> -d <dims to use> -k <k value> -r <resolution> -a <cluster algorithm> -m <cluster tmp file location> -o <output Seurat object in .rds format> -t <clusters in txt format>

Accessory scripts

Get a random set of genes

seurat-get-random-genes.R <Seurat object in .rds format> <output text file> <ngenes>

Convert formats

Multiple format conversion through Seurat 3:

• Possible input formats:
  • Seurat
  • AnnData (versions contemporary to Seurat 3).
  • Loom (versions contemporary to Seurat 3)
  • SingleCellExperiment
• Possible output formats:
  • Seurat
  • Loom
  • SingleCellExperiment

seurat-convert.R -i inputfile.rds --input-format seurat -o output.loom --output-format loom

This functionality should be used with care, as some elements of the objects can be lost in some conversions. When converting to and from Loom, be careful about table headers that might have offending characters for R tables.

Export to CellBrowser

Exports a Seurat RDS object and an (optional) markers file to a format that can be read by UCSC CellBrowser:

seurat-export-cellbrowser.R -i <Seurat object in .rds format> [-m markers.csv] -o <directory_for_output>