Single Cell Integration in Seurat v3.0
In our recent preprint, we introduce new methods into Seurat for single cell data integration. These include updated methods to integrate (or ‘assemble’) datasets into a common reference, as well as to transfer information from reference to query datasets.
Alongside this new functionality, we have made significant improvements to the Seurat object, accessor functions, and plotting library. In particular, we have upgraded the Seurat object to flexibly store multiple data types (‘assays’), from the same cells, and to allow the user to easily switch between them. We are currently preparing a full release of Seurat 3.0 on CRAN, with updated documentation, tutorials, and vignettes.
Here, we provide a pre-release of Seurat v3, with a brief vignette to enable users to explore our new methods, and to test them on their own datasets. While additional documentation and examples are forthcoming, please check out the following resources:
- Installation instructions (through the devtools package) for Seurat v3.0 pre-release
- Vignette: Integration and classification of human pancreatic islet cells
- Transitioning from Seurat v2 to Seurat v3:
- Command ‘cheat sheet’, a brief introduction to new commands, data accessors, visualization, and multiple assays in Seurat v3.0 as well as a translation guide between Seurat v2 and v3
- PBMC 3k guided clustering, a version of our PBMC 3k tutorial rewritten with Seurat v3 commands. This vignette is designed to reproduce the results of the Seurat v2 tutorial with v3 commands.
Seurat is an R package designed for QC, analysis, and exploration of single cell RNA-seq data. Seurat aims to enable users to identify and interpret sources of heterogeneity from single cell transcriptomic measurements, and to integrate diverse types of single cell data.
If you use Seurat v2.0 or above in your research, please considering citing Butler et al., Nature Biotechnology 2018. Seurat features three recently developed computational methods for single cell analysis:
- Unsupervised clustering and discovery of cell types and states (Macosko, Basu, Satija et al., Cell, 2015)
- Updated approach: Combining dimensional reduction with graph-based clustering
- Tutorial: Unsupervised identification of immune cell types and biomarkers from 2,700 PMBCs (10X Chromium)
- Spatial reconstruction of single cell data (Satija*, Farrell* et al., Nature Biotechnology, 2015)
- Integrates single cell RNA-seq with in situ reference data to infer cellular spatial localization from gene expression.
- Tutorial: Inferring spatial localization of single cells during Zebrafish embryogenesis
- Integrated analysis of single cell RNA-seq across conditions, technologies, and species (Butler et al., Nature Biotechnology, 2018)
- Unsupervised ‘alignment’ of shared cell types between single cell datasets, based on common sources of variation
- Tutorial: Integrating stimulated vs. control PBMC datasets to learn cell-type specific responses
All methods emphasize clear, attractive, and interpretable visualizations, and were designed to be easily used by both dry-lab and wet-lab researchers.
Seurat is developed and maintained by the Satija lab, in particular by Andrew Butler, Paul Hoffman, Christoph Hafemeister, and Shiwei Zheng, and is released under the GNU Public License (GPL 3.0). We are also grateful for significant ideas and code from Jeff Farrell, Karthik Shekhar, and other generous contributors.
March 23, 2018 Version 2.3 released
- Improvements for speed and memory efficiency
- New vignette for analyzing ~250,000 cells from the Microwell-seq Mouse Cell Atlas dataset
January 10, 2018: Version 2.2 released
- Support for multiple-dataset alignment
- New methods for evaluating alignment performance
October 16, 2017: Version 2.1 released
- Support for multimodal single cell data
- Support for MAST and DESeq2 packages for differential expression testing
July 26, 2017: Version 2.0 released
- Preprint published for integrated analysis of scRNA-seq datasets
- New methods for dataset integration, visualization, and exploration
- Significant restructuring of codebase to emphasize clarity and clear documentation
October 4, 2016: Version 1.4 released
- Added methods for negative binomial regression and differential expression testing for UMI count data
- New ways to merge and downsample Seurat objects
August 22, 2016: Version 1.3 released
- Improved clustering approach - see FAQ for details
- All functions support sparse matrices
- Methods for removing unwanted sources of variation
- Consistent function names
- Updated visualizations
May 21, 2015: Drop-Seq manuscript published. Version 1.2 released
- Added support for spectral t-SNE (non-linear dimensional reduction), and density clustering
- New visualizations - including pcHeatmap, dot.plot, and feature.plot
- Expanded package documentation, reduced import package burden
- Seurat code is now hosted on GitHub, enables easy install through devtools package
- Small bug fixes
April 13, 2015: Spatial mapping manuscript published. Version 1.1 released