### SynGeNet (Synergy from gene expression and network mining) ###

### OVERVIEW ###

SynGeNet R package implements a drug combination prediction method based on user-provided disease network data, drug target data and drug gene expression data. 


### SYSTEM REQUIREMENTS ###

The SynGeNet R code requires only a standard computer with enough RAM to support operations defined by a user. A computer with 8 GB of RAM will be sufficient.
We recommend installing the "msgsteiner" belief propagation algorithm dependency using a Linux operating system.
SynGeNet R code can be implemented on Mac and Windows operating systems. 


### INSTALLATION ###

Typical installation time on a computer with 8 GB of RAM requires 5 minutes. 

Required user provided data:
a) Fold-change gene expression for disease state of interest (co1= genes, col2= fc value), 'fc_filename.txt'; Example gene expression fold change data files included in GeneExpression folder for differentially expressed genes of four major genotypes of melanoma tumors from TCGA patient data

b) Root gene list for disease state of interest, 'root_filename.txt'; Example root gene files included in RootGene folder for co-mutated genes lists of four major genotypes of melanoma tumors from TCGA patient data

Note: Demo root gene list and gene expression data is included. 

Additional data files and code dependencies: 
a) SynGeNet functions included in this package: "subUserFunctionsVa2.R"

b) BioGRID interaction database: "BIOGRID-ALL-3.4.130.tab2-Symbol.txt". Please download "BIOGRID-ALL-3.4.130.mitab.zip" from: https://downloads.thebiogrid.org/BioGRID/Release-Archive/BIOGRID-3.4.130/, and convert the data to a 2-column table, where 1st and 2nd columns contain the Gene symbols of protein-protein interactions

c) Belief propagation program contained in "msgsteiner" folder:
1. Download the code from: http://areeweb.polito.it/ricerca/cmp/code (version: msgsteiner.tgz 6.84 KB)
2. Install 'boost': brew install boost
3. 'make' command (in the README) to compile the program
 note: Mac OS may not support this program. We recommend installing on a Linux system.

d) Connectivity mapping drug-induced gene expression data, example included: "melanoma_fdadrugs_n633.txt"

e) DrugBank drug-target interaction data: "uniprot links.csv"
download the drugbank drug-target data from: https://www.drugbank.ca/releases/latest#external-links

f) Stitch db drug-target interaction data: "9606.protein_chemical.links.detailed.v4.0.tsv"
download the stictch drug target data from: http://stitch.embl.de/cgi/download.pl?UserId=I5NZdVgNfHhb&sessionId=NEYKoy3UWESS&species_text=Homo+sapiens
select the 'homo sapiens', and download the protein-chemical file.

g) Download the GSEA function from https://rdrr.io/bioc/EnrichmentBrowser/src/R/GSEA.1.0.R, and copy and paste this function into the 'subUserFunctionsVa2.R' file included in this package: GSEA.EnrichmentScore2 <- function(gene.list, gene.set, weighted.score.type = 0, correl.vector = NULL))


### DEMO ###
The example workflow included in this package replicates drug combinations predicted for different four different melanoma genomic subtypes (BRAF, NRAS, NF1 and triple wild-type) ranked by calculated synergy score. The expected run time using the provided demo datasets is approximately 1 hour.