Met2Graph implements a flexible process flow to build graphs starting from a genome-scale metabolic model and can be easily integrated with user-customized functions. It allows the creation of three different types of graphs, based on selection of nodes, edges, and attributes. This document provides a general overview of the functionalities presented in the Met2Graph package. Below, a tutorial on how to use the package functions and get different metabolic graphs is provided.

To report bugs and arising issues, please visit https://github.com/cds-group/Met2Graph

Met2Graph needs the package sybilSBML, which in turn depends on libSBML to import and read SBML files. Running functions could fail if this prerequisite library is not available. Please read through the instructions provided at https://github.com/cran/sybilSBML/blob/master/inst/INSTALL.

The devtools package is needed to install Met2Graph directly from github.

Package devtools is available at CRAN. For Windows, this seems to depend on having Rtools for Windows installed. You can download and install this from: http://cran.r-project.org/bin/windows/Rtools/

To install R package devtools call:

    install.packages("devtools")

In R console, type:

    library(devtools)
    install_github(repo="Met2Graph", username="cds-group")

First, load the library.

library(Met2Graph)

The metabolites are the nodes, connected by reactions and, if present, by the relative catalyzing enzymes.

Two metabolites are connected if, in one or multiple reactions, one is a substrate and the other one is a product.

The edges can be weighted by the expression values of the corresponding enzymes.

The example below creates a Metabolites-based graph for each sample of expression data of 3 kidney tumor samples downloaded from the Genomic Data Commons data portal (https://portal.gdc.cancer.gov) in the form of FPKM (fragments per kilobase per million reads mapped) normalized read counts (the directory containing the expression files is indicated by "exprDir" argument). The network is created from the metabolic model of kidney tissue downloaded from https://metabolicatlas.org/ in SBML format ("infile").

The parameter "catalyzed"" is TRUE, and only reactions with associated enzymes are kept.

Recurring metabolites (e.g., ATP, CO2, NADP) can be removed to avoid unrealistic paths by setting the parameter "rmMets"==TRUE and using the list of these compounds contained in the data of the package by default or one provided by the user through the parameter "recMets".

The gene-protein-reaction (GPR) associations are automatically extracted from the model. If not present, those extracted from the human generic GEM are used. Below, Human generic GEM gpr rules downloaded from https://metabolicatlas.org/.

The resulting graph can be obtained with multiple edges corresponding to the multiple enzymes connecting the two metabolites or can be simplified to single edges. The choice is indicated by the parameter "simpl", where a logical value is required.

Several methods of simplification are proposed, and can be indicated through the parameter "GPRparse". For this parameter three options are possible:

1. "meanSum": for each edge, expression values of enzymes acting in the same reaction are averaged, and these averages are then summed up for enzymes acting in different reactions;
2. "minMax": according to the GPR associations, enzymes catalyzing the same reaction are associated by "AND" or "OR" logical relationships based on complex or mutual exclusive activity, respectively. For each edge, the weight is given by the minimum expression value of enzymes related by "AND" and the maximum value for "OR". Again, the values of enzymes acting in different reactions are summed up;
3. "minSum": according to the GPR associations, enzymes catalyzing the same reaction are associated by "AND" or "OR" logical relationships based on complex or mutual exclusive activity respectively. For each edge, the weight is given by the minimum expression value of enzymes related by "AND" and the summed value for "OR". Finally, the values of enzymes acting in different reactions are summed up.

In the example multiple edges are simplified by "meanSum" (simpl=TRUE, GPRparse="meanSum").

The parameter "add_rev_rxn" is by default set to FALSE, otherwise the edges will be assigned considering both directions of reversible reactions.

The resulting graph can be built directed or not. Given the rules behind the definition of connections, the networks is naturally directed and the "directed" parameter is by default set to TRUE. Nonetheless the user can choose to not consider this aspect, setting the "directed" parameter to FALSE. Below, the obtained graphs are weighted and directed.

"outDir" indicates the directory to store the networks and "outFormat" the format in which save the files (see igraph documentation for the other formats).

infile <- system.file("extdata", "kidney.xml", package = "Met2Graph")
exprDir <- system.file("extdata/expression/", package = "Met2Graph")
outDir <- system.file("extdata/output/", package = "Met2Graph")
Met2MetGraph(infile, catalyzed = TRUE, rmMets = TRUE, exprDir=exprDir, simpl = TRUE , GPRparse = "meanSum", outDir= outDir, outFormat = "ncol")

Output files:

1. A graph for each sample in the desired format ("ncol" in this case).

2. In case of simplification a matrix with simplified weights of all samples for each metabolites couple.

If the simplification of edges is not applied, a matrix with not simplified expression values for each enzyme, corresponding metabolites couple and reactions is provided.

Reactions are the nodes connected by edges represented by metabolites.

Two reactions are connected if one produces a metabolite consumed by the other one.

In the example below, the metabolic model of kidney, as above, is given by "infile" argument.

Recurring metabolites (e.g. ATP, CO2, NADP) can be removed to avoid unrealistic paths setting the parameter "rmMets"==TRUE and using the list of these compounds contained in the data of the package by default or one provided by the user through parameter "recMets".

The parameter "add_rev_rxn" is by default set to FALSE, otherwise the edges will be assigned considering both directions of reversible reactions.

"outDir" indicates the directory to store the networks and "outFormat" the format in which save the files (see igraph documentation for the other formats).

The obtained graphs are unweighted and directed.

infile <- system.file("extdata", "kidney.xml", package = "Met2Graph")
outDir <- system.file("extdata/output/", package = "Met2Graph")
Met2RxnGraph(infile, rmMets = TRUE, outDir=outDir, outFormat = "ncol")

Output files:

1. Graph in the desired format ("ncol" in this case) and a tabular file containing reaction nodes and metabolite edges.

Enzymes are the nodes connected by edges represented by metabolites.

Two enzymes are connected if they catalyze two reactions which produce and consume a metabolite, respectively.

Enzymes for each reaction are annotated according to GPR rules. Enzymes related by AND are complexes and considered as a single node, while OR relationships are split as different nodes. The GPR associations are needed. They are automatically extracted form the model. If not present in the model, the one from the human generic GEM is used.

Recurring metabolites (e.g., ATP, CO2, NADP) can be removed to avoid unrealistic paths by setting the parameter "rmMets"==TRUE and using the list of these compounds contained in the data of the package by default or one provided by the user through the parameter "recMets".

The parameter "add_rev_rxn" is by default set to FALSE, otherwise the edges will be assigned considering both directions of reversible reactions.

The resulting graph can be built directed or not. Given the rules behind the definition of connections, the networks is naturally directed and the "directed" parameter is by default set to TRUE. Nonetheless the user can choose to not consider this aspect, setting the "directed" parameter to FALSE. Below, the obtained graphs are unweighted and directed.

"outDir" indicates the directory to store the networks and "outFormat" the format in which save the files (see igraph documentation for the other formats).

infile <- system.file("extdata", "kidney.xml", package = "Met2Graph")
outDir <- system.file("extdata/output/", package = "Met2Graph")
Met2EnzGraph(infile, rmMets=TRUE, outDir=outDir, outFormat="ncol")

Output files:

1. Graph in the desired format ("ncol" in this case) and a tabular file containing enzyme nodes and metabolite edges.

Plot_graphs plots the graphs to pdf. Input directory containing the graphs in the specific "format" needs to be indicated ("inDir"), as well as the path where to save the pdf ("outDir"). Additional plotting parameters can be set (In the example below, e.g., "layout", "vertex.size", "edge.arrow.size", "vertex.label"). See igraph.plotting for the complete list.

inDir <- system.file("extdata/output/MetGraphs/", package = "Met2Graph")
outDir <- system.file("extdata/output/", package = "Met2Graph")
Plot_graphs(inDir=inDir, outDir=outDir, format="ncol", layout=layout_with_graphopt, vertex.size= 1,edge.arrow.size=0.1,vertex.label=NA)

To give the possibility to explore the model and have a look to the list of reactants and products for each reaction used to create the networks above, the function "MetExtract" is available. It imports and reads a metabolic model in SBML format to get lists of reactants and products for each reaction. In the example below, the metabolic model of kidney tissue downloaded from https://metabolicatlas.org/ is imported. The output objects, consisting of model and two dataframes corresponding to reactants and products, are stored in a list object.

out <- MetExtract(system.file("extdata", "kidney.xml", package = "Met2Graph"))

If you use the Met2Graph package, please cite the following paper: Granata, I., Manipur, I., Giordano, M., Maddalena, L., Guarracino, M.R. TumorMet: A repository of tumor metabolic networks derived from context-specific Genome-Scale Metabolic Models, Scientific Data, submitted (2022)