Skip to content

Latest commit

 

History

History
167 lines (117 loc) · 7.78 KB

README.md

File metadata and controls

167 lines (117 loc) · 7.78 KB
Raw

scater: single-cell analysis toolkit for expression with R

Linux Build Status Coverage Status Windows Build Status

This package contains tools for the analysis of single-cell gene expression data using the statistical software R. The package places an emphasis on tools for quality control, visualisation and pre-processing of data before further downstream analysis.

We hope that scater fills a useful niche between raw RNA-sequencing count or transcripts-per-million data and more focused downstream modelling tools.

Briefly, scater enables the following:

  1. Automated computation of QC metrics
  2. Transcript quantification from read data with pseudo-alignment
  3. Rich visualisations for exploratory analysis
  4. Seamless integration into the Bioconductor universe using the SingleCellExperiment class
  5. Simple normalisation methods and tight integration with the scran package.

See below for information about installation, getting started and highlights of the package.

Installation

Installation from Bioconductor (recommended)

The scater package has been accepted into Bioconductor! Thus, the most reliable way to install the package is to use the usual Bioconductor method:

## try http:// if https:// URLs are not supported
if (!requireNamespace("BiocManager", quietly=TRUE))
    install.packages("BiocManager")
BiocManager::install("scater")

The scater package has been available as a "release" version in the Bioconductor since April 2016. The release version of scater works with the release version of R and Bioconductor, and development will continue in the devel version of the package on Bioconductor. Future releases will follow the regular Bioconductor release cycle.

Using the most recent version of R is strongly recommended (R 3.4 at the time of writing). Effort has been made to ensure the package works with R >3.0, but the package has not been tested with R <3.1.1.

There are several other packages from CRAN and Bioconductor that scater uses; installing through Bioconductor will install these packages as well.

The following optional packages are not strictly required but enhance the functionality of scater:

install.packages(c("cowplot", "cluster", "mvoutlier", "parallel", "Rtsne"))
BiocManager::install(c("destiny", "monocle"))

You might also like to install dplyr for convenient data manipulation:

install.packages("dplyr")

Getting started

The best place to start is the vignette. From inside an R session, load scater and then browse the vignettes:

library(scater)
browseVignettes("scater")

There is a detailed HTML document available that introduces the main features and functionality of scater.

The step-by-step workflow offers further examples of using scater and scran for low-level analysis of scRNA-seq data.

scater workflow

The diagram below provised an overview of the pre-processing and QC workflow possible in scater, listing the functions that can be used at various stages. A first place to start is with the newSCESet function, which will allow you to create a data object for use with scater.

Diagram outlining the scater workflow

Highlights

The scater package allows you to do some neat things relatively quickly. Some highlights are shown below with example code and screenshots.

  1. Automated computation of QC metrics
  2. Transcript quantification from read data with pseudo-alignment approaches
  3. Data format standardisation
  4. Rich visualisations for QC and exploratory analysis
  5. Seamless integration into the Bioconductor universe
  6. Simple normalisation methods

For details of how to use these functions, please consult the vignette and package documentation. The plots shown use the example data included with the package (for which there is no interesting structure) and as shown require only one or two lines of code to generate.

Automatic computation of QC metrics

Use the calculateQCMetrics function to compute many metrics useful for gene/transcript-level and cell-level QC. Metrics computed include number of genes expressed per cell, percentage of expression from control genes (e.g. ERCC spike-ins) and many more.

Transcript quantification with kallisto or Salmon

The runKallisto and runSalmon functions provides wrappers to the kallisto and 'Salmon' software for quantifying transcript abundance from FASTQ files using a "pseudo-alignment"" or "lightweight alignment" approaches. These new approaches are extremely fast while retaining accuracy. With readKallistoResults and readSalmonResults, transcript quantities can be read into a data object in R.

Plotting functions

Default plotScater for an SCESet object gives cumulative expression for the most-expressed features (genes or transcripts)

The plotTSNE function produces a t-distributed stochastic neighbour embedding plot for the cells.

The plotPCA function produces a principal components analysis plot for the cells.

The plotDiffusionMap function produces a diffusion map plot for the cells.

The plotExpression function plots the expression values for a selection of features.

The plotQC function produces a variety of QC plots useful for diagnostics and feature and cell filtering. It can be used to plot the most highly-expressed genes (or features) in the data set or create density plots to assess the relative importance of explanatory variables, as well as many other visualisations useful for QC.

The plotPhenoData function plots two phenotype metadata variables (such as QC metrics).

See also plotFeatureData to plot feature (gene) metadata variables, including QC metrics.

Plus many, many more possibilities. Please consult the vignette and documentation for details.

Acknowledgements and disclaimer

The package leans heavily on previously published work and packages, namely edgeR and limma. The SingleCellExperiment class from the SingleCellExperiment package (new for Bioconductor 3.6+) provides a modern data structure to support single-cell analyses. scater has adopted this data structure from Bioconductor 3.6; wide adoption across Bioconductor will streamline analysis workflows using multiple packages.

The scater sticker is licensed under Creative Commons Attribution CC-BY. Feel free to share and adapt, but don't forget to credit the author. Skateboard icon made by Nikita Golubev from Flaticon is licensed by Creative Commons BY 3.0.

We hope the scater package makes your life easier when analysing single-cell RNA_seq data. Please do try it and contact us with bug reports, feedback, feature requests, questions and suggestions to improve the package.

Davis McCarthy, September 2017 (on behalf of scater authors and contributors)