Single Cell Pathway Analysis

On this page

  1. Recent updates to SCPA
  2. A brief overview of SCPA
  3. Package installation
  4. Links to tutorials
  5. Submitting issues/comments
  6. Improvements/changes/updates to SCPA

Recent updates

  • Compatability with Seurat v5 objects to use the layer system in seurat_extract() and compare_seurat() (v1.6). SCPA automatically detects Seurat object version
  • Ability to specify multiple gmt files as your pathway input (v1.5.2)
  • Parallel processing implemented to speed up analyses (v1.5.0)
  • Much more efficient usage of memory, so this shouldn’t be limiting (v1.3.0)

About SCPA

SCPA is a method for pathway analysis in single cell RNA-seq data. It’s a different approach to pathway analysis that defines pathway activity as a change in multivariate distribution of a given pathway across conditions, rather than enrichment or over representation of genes.

This approach allows for a number of benefits over current methods:

  1. Multivariate distribution testing allows for the identification of pathways that show enrichment in a given population AND also pathways that show transcriptional change independent of enrichment. You essentially get the best of both worlds, as pathways with changes in multivariate distribution (high qval) but no overall enrichment (low fold change) are still interestingly different pathways, as we show in our paper. For more on this, see our SCPA interpretation page

  2. SCPA allows for multisample testing, so you can compare multiple conditions simultaneously e.g. compare across 3 time points, or across multiple phases of a pseuodotime trajectory. This means you can assess pathway activity through multiple stimulation phases, or across cell differentiation

SCPA can be applied directly on Seurat and SCE objects, as well as manually subsetted expression matrices. You just need to supply this along with your pathways that you want to analyse.

To see the stats behind SCPA, you can see our paper in JASA here

Our paper introducing SCPA and demonstrating its use on a T cell scRNA-seq dataset is published in Cell Reports here

Installation

You can install SCPA by running:

# install.packages("devtools")
devtools::install_version("crossmatch", version = "1.3.1", repos = "http://cran.us.r-project.org")
devtools::install_version("multicross", version = "2.1.0", repos = "http://cran.us.r-project.org")
devtools::install_github("jackbibby1/SCPA")

Installation problems

If you’re running into installation errors, you’ll likely need to manually install the dependencies that are mentioned in the error. For example, see these issues here and here

Tutorials

If you’re viewing this page on GitHub, the SCPA webpage with all the documentation and tutorials is here

We have various examples and walkthroughs, including:

  • A generic quick start tutorial on a basic scRNA-seq dataset
  • A tutorial on how to get and use gene sets with SCPA
  • An outline for speeding up SCPA with parallel processing
  • A tutorial for more detailed two group comparison with a specific scRNA-seq data set
  • A tutorial on how to use SCPA directly within a Seurat or SingleCellExperiment object
  • A tutorial on visualising SCPA output
  • A tutorial for multisample SCPA, comparing pathways across a scRNA-seq T cell pseudotime trajectory
  • A tutorial of a systematic analysis of many cell types across multiple tissues
  • A tutorial for a systems level analysis of many cells types in disease (COVID-19)

Issues

To report any issues or submit comments please use: https://github.com/jackbibby1/SCPA/issues

Changelog

Any new features or alterations to SCPA can be found here