CellWalkR is updated to Version 2! This version combines the functionality of the CellWalker model with new features implemented in CellWalker2. As inputs, CellWalkerR takes cell type labels (lineages, states, etc; defined by their marker genes) and count data: a gene by cell matrix from scRNASeq data and/or a peak by cell matrix from scATACSeq data. Count matrices can be computed using a variety of upstream single-cell quantification software tools. Optionally, users may provide genome coordinates for sets of annotations they wish to map to cell types. Examples of annotations are bulk-derived regulatory elements, sequence motifs, genetic variants, or gene sets. CellWalkR builds a graph where the nodes are cells, cell types, and (if provided) annotations. Then, CellWalkR uses a graph diffusion method to annotate cells, compare cell type labels, and assign cell type-specificity to annotations.
The original CellWalker model integrates single-cell open chromatin (scATAC-seq) data with cell type labels and (optionally) bulk epigenetic data to annotate cells, compare cell type labels, and probabilitically assign cell type labels to annotations. CellWalker2 extends this functionality with many new features, including:
- integrates different modalities of single-cell data (scATAC-seq, scRNA-seq, multiomic),
- enables integrative modeling of cells measured in different experiments,
- incorporates hierarchical relationships between cell type labels,
- compares cell type ontologies across contexts (conditions, species, datasets),
- provide permutation-based measures of statistical significance for cell-to-label, region-to-label, and label-to-label mappings.
Install CellWalkR for R using devtools as follows:
$ R
> install.packages("devtools")
> devtools::install_github("PFPrzytycki/CellWalkR@cellwalker2")For a guide to use CellWalker, see the provided readme and vignette.
For a guide to use CellWalker2, see the provided readme.
If you use CellWalkR please cite:
-
Przytycki, P.F., Pollard, K.S. “CellWalkR: An R Package for integrating and visualizing single-cell and bulk data to resolve regulatory elements.” Bioinformatics (2022). https://doi.org/10.1093/bioinformatics/btac150
-
Przytycki, P.F., Pollard, K.S. “CellWalker integrates single-cell and bulk data to resolve regulatory elements across cell types in complex tissues.” Genome Biology (2021). https://doi.org/10.1186/s13059-021-02279-1
CellWalkR can also be run on AWS which vastly simplifies the process of running on GPUs using TensorFlow. Using GPUs allows the code to run more than 15 times faster. For a guide to running CellWalkR on AWS using GPUs see this vignette.