#####################
### -- Updates -- ###
#####################

29/11/2022: 
An update to the GDC portal seems to have broken the TCGA2STAT library. We're working on a fix.
In the meanwhile, you can download the relevant datasets from our storage (see: NOTE2 below).
Note that our tool works with any manually downloaded and prepared data as well.

################################
### --- HOW TO RUN ViLoN --- ###
################################

With the code and files provided in this folder you can regenerate results for the THCA cancer discussed in the corresponding publication.
Please open each of the stepX... scripts in ascending order, and run the R code provided.
Step3 needs to be run via commandline with Python, as specified in the corresponding file.
Step4 for LRACluster can be found in the LRACluster/ folder, and needs to be run there.
Necessary folders will be generated on the fly.

NOTE: 
The respective files include in-line documentation, which explains what happens at each step, especially where what happens is not self-explanatory.

NOTE2: 
For your convenience, the TCGA datasets analysed in this manuscript are also stored under https://bicloud.boku.ac.at/index.php/s/jAfiWriPb2tGHHN
(pwd: nar_submission2022)
These files include both the molecular data and clinical information, stored as an R object.
You can download the data to a corresponding folder and skip step1_getData.ViLoN.R
This is to ensure that even if the TCGA2STAT tool becomes outdated, you still have access to the relevant datasets.

#############################
### --- PREREQUISITES --- ###
#############################

You will need to install the following libraries in order to be able to run the code

Python3: snakemake, listdir, os

R: TCGA2STAT,limma,edgeR,foreach,doMC,SNFtool,survival,coxphf,igraph,blockmodels

In case your R version does not support TCGA2STAT anymore, you can install TCGA2STAT and its dependencies by putting the below code into a tcga2stats-install.R script:

### code snippet start
repo<-"https://cloud.r-project.org/";

if (!require("BiocManager",quietly=TRUE)) {
    res<-try(install.packages("BiocManager",repos=repo,quiet=TRUE));
    if (inherits(res,'try-error')) {
        usrlib<-Sys.getenv("R_LIBS_USER")[1];
        cat("/== Installing into user library",usrlib,"\n");
        dir.create(usrlib,recursive=TRUE,showWarnings=FALSE)
        install.packages("BiocManager",lib=usrlib,repos=repo);
        .libPaths(c(usrlib,.libPaths()));
    }
}

BiocManager::install("CNTools");
install.packages("https://cran.r-project.org/src/contrib/Archive/TCGA2STAT/TCGA2STAT_1.2.tar.gz")
### code snippet end

...and then, in the same folder, from command line, calling:
R --vanilla <tcga2stats-install.R 2>&1 |tee tcga2stats-install.log

##################################
### --- COMPUTATIONAL TIME --- ###
##################################

The construction of the full network, starting from first building bipartite graphs, all the way to constructing the final patient-patient network, takes around 6h on a modern workstation (we used 36 cores).