This is a repository of data, code and analyses related to the paper "Personalization of Logical Models With Multi-Omics Data Allows Clinical Stratification of Patients". The paper can accessed here: https://www.frontiersin.org/articles/10.3389/fphys.2018.01965.
This repository can be cited with its own DOI:
A step-by step tutorial has been added. Please have a look at Tutorial_PROFILE.pdf and its instructions for a simple pipeline with similar steps but different models and data.
This set of files and scripts is supposed to be self-sufficient. Please download data and scripts and follow instructions
- Python version 3.0 or greater
- Python's package pandas
- Perl
- R
- MaBoSS requires: flex, bison, gcc and g++
In the present pipeline, two different datasets may be used (METABRIC or TCGA) and processed for further simulations with two different logical models, either a generic one (Fumia model, see paper) or a breast-specific one (Zanudo model, see paper).
The following instructions will use METABRIC data and Fumia model.
Patient-profiles generation and all its underlying computations (funcional effect inference, normalization and binarization) have been packed in a .Rmd file. Please run Fumia_META_profiles.Rmd (in Scripts/Profiles folder) with R to generate profiles (in Results/Profiles folder) and a corresponding .html report (in Scripts/Profiles folder)
rmarkdown::render("Fumia_META_profiles.Rmd", "html_document")
Patient-specific profiles are used to personalise patient-specific models through different methods based on node activity status, initial conditions or transition rates. Simulations are performed by MaBoSS software.
However, model personalisation requires to modify MaBoSS model files. The whole simulation process is packed in python script. Please note that depending on your operating system, you should choose different versions of that script (either MaBoSS_specific_Mac or MaBoSS_specific_Linux)
A minimal example of simulations can be found in the following shell script. This is an example of simulations using mutations and CNA information of METABRIC cohort as node activity status. Only values for model output nodes (dead-end nodes regulated by inhibitors/activators but not regulating any other node) are saved.
model=Fumia2013
sim_case = META_mutations_CNA_asMutants
python3 Scripts/Simulations/MaBoSS_specific_Mac.py $model 2 "Results/Simulations/results_"$sim_case".txt"
-s $sim_case -m "Results/Profiles/Fumia_META_mutCNA.csv"
This example is available in example1.sh, either in MacOS or Linux distribution versions.
Another example, computing the scores for all nodes of the model, is available in example2.sh, either in MacOS or Linux distribution versions. The loop structure is required to avoid an exponential increase of computation time.
Scripts were mostly designed by Jonas Béal (jonas dot beal at curie dot fr). See a complete list of authors in the corresponding paper.