DIALOGUE is a dimensionality reduction method that uses cross-cell-type associations to identify multicellular programs (MCPs) and map the cell transcriptome as a function of its environment. Given single-cell data, it combines penalized matrix decomposition with multilevel modeling to identify generalizable MCPs and examines their association with specific phenotypes of interest.
- R (tested in R version 3.4.0).
- R libraries: lme4, lmerTest, PMA, plyr, matrixStats, psych, stringi, RColorBrewer, unikn, reshape2, ggplot2, grid, beanplot
To install DIALOGUE you can either use devtools::install_github("DIALOGUE",your_user_name) or just download the package and use devtools::install("DIALOGUE")
See the tutorial for more details.
All you need for the input is the single-cell transcriptomes of different cell types, usually together with some more compact representation (e.g., PCs). The output will be multicellular programs (MCPs) of co-regulated genes across the different cell types. Each MCP consists of cell-type-specific gene subsets.
For specific cell-cell "interactions" you can run the pairwise version, using the data of two cell types of interest as input. DIALOGUE can also identify MCPs that span multiple cell types (see Jerby-Arnon and Regev bioRxiv 2020 for examples on 5 and 6 cell types).
Jerby-Arnon and Regev Mapping multicellular programs from single-cell transcriptomes.