Deconvolution and Multitask Regression-Based Imputation • DURIAN

DURIAN (Deconvolution and mUltitask-Regression-based ImputAtioN) is a hybrid, iterative method for imputation of drop-out in single-cell expression data. Automatic recovery of missing single-cell reads is supervised by deconvoluted bulk expression data.

For examples of pipelines involving DURIAN, users should consult the scripts used to generate the benchmarking and figures for our paper.

Get Started

Installation: see below
Basic usage of DURIAN running and visualizing imputation of Human single-cell data in vignette("basic_usage").
Selection of the DURIAN deconvolution module, including R/C++ and Julia-based modules vignette("module_selection").
Use DURIAN on Seurat objects directly, and perform cell signaling analysis in CellChat in vignette("signaling").

Installation

Dependencies

ENV["JULIA_PROJECT"]=pwd()
ENV["R_HOME"]="/usr/local/lib/R"
using Pkg
Pkg.add("Distributed")
Pkg.add("RCall")
Pkg.build("RCall")
using RCall

Pkg.add(url="https://github.com/mkarikom/DistributedStwdLDA.jl.git")

# Install the latest version and dependencies from github
library(devtools)
devtools::install_github("mkarikom/MuSiC@biobaseImport")
devtools::install_github("mkarikom/mtSCRABBLE")
install.packages("foreach")

Sys.setenv(JULIA_HOME="/usr/local/julia-1.6.6/bin")

install.packages("JuliaCall")
library(JuliaCall)
julia_setup(JULIA_HOME = Sys.getenv("JULIA_HOME"),verbose=TRUE,rebuild = TRUE,install=FALSE)

Package

install_github("mkarikom/DURIAN")