piTracer - Automatic reconstruction of molecular cascades for the identification of synergistic drug targets
The piTracer framework reconstructs molecular cascades between two genes, two metabolites, or a gene and a metabolite based on an underlying multi-omics pathway network. Its drug ranking functionality can identify synthetic lethal candidates based on a differential metabolomics experiment after single agent treatment.
Gomari D, Achkar I, Benedetti E, Tabling J, Halama A, Krumsiek J. "piTracer - Automatic reconstruction of molecular cascades for the identification of synergistic drug targets". biorXiv [link]
A hosted version of the Shiny app version of piTracer is can be found here.
Clone this repository to your local machine
git clone https://github.com/krumsieklab/piTracer.git
piTracer was tested on R 4.2.1 with the corresponding latest packages as of early 2023. We recommend using the package versions stored using renv. To set this up, open the piTracer project file piTracer.proj in RStudio, and run:
renv::restore()
The app requires the precalculated binary files from the precalculated_data folder to run. To clone those files as part of the repository, you need to have git-lfs set up. If you don't have access to git-lfs, you will need to download the files manually.
Run run_piTracer.R or excute to launch the Shiny app in your local browser.
The script /src/gene_ranking/execute_generanking.R runs the ranking algorithms programmatically based on an input spreadsheet.
| Folder | Description |
|---|---|
src/global.R |
used by various scripts to set up the tracing engine, Shiny app modules, and precalculated data files |
src/gene_ranking/ |
contains the drug combination prediction functionality |
src/shiny_app/ |
contains UI and server modules of the Shiny app |
src/tracing/ |
contains scripts that run the core tracing steps |