README.md

In cbg-ethz/mnem: Mixture Nested Effects Models

M&NEM

Single cell RNA-seq data sets from pooled CrispR screens provide the possibility to analyzse hete rogeneous cell populations. We extended the original Nested Effects Models (NEM) to Mixture Nested Effects Models (M&NEM) to simulataneously identify several causal signalling graphs and corresponding subpopulations of cells. The final result will be a soft clustering of the perturbed cells and a causal signalling graph, which describes the interactions of the perturbed genens for each cluster of cells.

Install:

if (!requireNamespace("BiocManager", quietly = TRUE))
    install.packages("BiocManager")

BiocManager::install("mnem")

Most recent (devel) version:

install.packages("devtools")

library(devtools)

install_github("cbg-ethz/mnem")

library(mnem)

Small toy example with five S-genes and 1000 simulated cells. Each S-gene has two E-genes. The two components have weights 40 and 60 percent. The simulated data set consists of log ratios for effects (1) and no effects (-1). We add Gaussian noise with mean 0 and standard deviation 1. We learn an optimum with components set to two and ten random starts for the EM algorithm.

sim <- simData(Sgenes = 5, Egenes = 2, Nems = 2, mw = c(0.4,0.6))
data <- (sim$data - 0.5)/0.5
data <- data + rnorm(length(data), 0, 1)
result <- mnem(data, k = 2, starts = 10)
plot(result)

For the reproduction of the publication see the scripts in the other directory.

References:

Pirkl, M., Beerenwinkel, N.; Single cell network analysis with a mixture of Nested Effects Models, Bioinformatics, Volume 34, Issue 17, 1 September 2018, Pages i964-i971, https://doi.org/10.1093/bioinformatics/bty602.

cbg-ethz/mnem documentation built on Nov. 7, 2024, 7:35 p.m.