The pharmacogenomic assessment of molecular epithelial-mesenchymal transition signatures reveals drug susceptibilities in cancer cell lines
This repository contains the analysis steps for assessing epithelial-mesenchymal transition scores as predictive biomarkers (emtpb) for drug response in cancer cell lines.
Most of the required software is available in a docker container. Furthermore, nextflow and charliecloud need to be installed for running the benchmarks.
Pull the repository with
# Clone the repository
git clone https://github.com/mendenlab/emtpb.git
The required software for running the analysis is available on a docker container. It can be either pulled from dockerhub by typing docker pull aljoshoh/emtpb
(recommended), or it can be rebuilt from scratch:
# Build and export container
cd emtpb/environment
docker build --no-cache -f Dockerfile -t "aljoshoh/emtpb" .
docker export $(docker create aljoshoh/emtpb) | gzip -c > metadata/emtpb.tar.gz
After having the container available, you can run the image by
docker run --rm -it -p 3838:3838 aljoshoh/emtpb /bin/bash
Within the image can run the analysis steps in scripts/
. You can also run them within rstudio, which you can access by typing
/bin/start_rstudio 3838
while accessing the session in http://localhost:3838
with rstudio
as username and the generated password.
The ordered analysis pipeline in scripts/
can be run within the created environment. Make sure to adjust your system paths in scripts/config.yaml
. While each of the R scripts only needs to be run once, the benchmarks in scripts/02_benchmark_*
will need to be run in batch mode using nextflow (with adjusted paths). Then, un each of the two benchmarks by:
# Run 1
cd nf
nextflow run train_chc.nf --file=scripts/02_benchmark_PANCAN.ipynb --amount=313600 --desc=exp3 -profile slurm -bg -with-trace -ansi-log false --Xms500M --Xmx2G > nf.log
# Run 2
cd nf_2
nextflow run train_chc.nf --file=scripts/02_benchmark_CI.ipynb --amount=156800 --desc=exp6 -profile slurm -bg -with-trace -ansi-log false --Xms500M --Xmx2G > nf.log
The figures for the manuscript can be reproduced by running paper/figures_02.Rmd
.
This repository refers to our latest manuscript (citation pending).