Transportome Profiler analysis

This repository contains the code for the analysis on the expression profile of the transportome in Cancer based on the MTP-DB.

[!NOTE] Read the preprint here: Profiling the Expression of Transportome Genes in cancer: A systematic approach

It's potentially out of date.

The project follows the Kerblam! standard.

Running the analysis

You can run the analysis pipelines with Kerblam! and docker:

# Clone the repo
git clone git@github.com:TCP-Lab/transportome_profiler.git
cd ./transportome_profiler

kerblam data fetch # Fetch the input data not present in the repository
kerblam run <pipeline>

Kerblam! will build docker containers and run the analysis locally. To run without docker, read below.

Pipelines

The project currently encompasses the following pipelines:

• heatmaps: Create large heatmaps from the expression matrices by using GSEA on computed gene rankings, testing all possible gene lists that can be made from the MTP-DB.
  • The test profile makes this pipeline much faster by running on smaller (i.e. sampled) input data (~75% reduction in sample number, only 5000 random genes).

Running locally without docker

You need some requirements to be installed before you can run the analysis locally:

• R version 4.3.0+.
  • Install R requirements with ./src/helper_scripts/install_r_pkgs.R.
• Python version 3.11+.
  • Install python requirements with pip install -r ./src/requirements.txt.
• The jq utility (that you can find here).
• The xsv program, required by metasplit (sudo pacman -Syu xsv on Arch, not packaged by Debian, but this guide might be useful. If you have cargo installed, you can simply run cargo install xsv).
• Follow the extra installation guide for generanker (namely installing fast-cohen)
• The xls2csv utility (on arch yay -Syu perl-xls2csv)
• A series of R packages that can be installed with Rscript ./src/helper_scripts/install_R_pkgs.R
• Quite a bit of RAM (some steps require > 50 Gb of RAM) and time. Override N_THREADS (with export N_THREADS=...) to run with less threads.

If you have all the requirements, you can simply:

kerblam run <pipeline> --local

[!IMPORTANT] The manuscript for this project is also available online in this repository.