DWLS

Forked from https://bitbucket.org/yuanlab/dwls/src/master and modified for speed.

Installation

You can install the development version of DWLS like so:

remotes::install_github('alexvpickering/DWLS')

Example

DWLS requires 3 inputs:

cluster labels and marker genes
single-cell counts
bulk counts

Marker genes in the appropriate format can be obtained from single-cell counts in a variety of ways. For example

library(DWLS)

library(Seurat)
data(pbmc_small)
id <- Idents(pbmc_small)
scdata <- pbmc_small[['RNA']]@counts

markers <- DEAnalysisPresto(scdata, id)

The next step is to build the single-cell signature matrix:

sig <- buildSignatureMatrix(scdata, id, markers)

Next run deconvolution:

# example bulk signature
bulk <- rowSums(scdata)

# trim to common genes
tr <- trimData(sig, bulk)

# run deconvolution
res <- solveDampenedWLS(tr$sig, tr$bulk)

# parallel implementation for multiple bulk signatures
bulk2 <- cbind(bulk, bulk)
tr2 <- trimData(sig, bulk2)
res2 <- solveParallel(tr2$sig, tr2$bulk)