ICON-GEMs
Authors: Thummarat Paklao
, Apichat Suratanee
, and Kitiporn Plaimas
, Apichat Suratanee, and Kitiporn PlaimasICONGEMs (Integration of CO-expression Network into GEnome scale Metabolic models) [1] - an approach to integrating gene co-expression networks into FBA models, allowing for more accurate determination of flux distribution and functional pathways. By constructing a comprehensive gene co-expression network, we obtained a global perspective on the cell's mechanism. Using quadratic programming, we optimized the alignment between pairs of reaction fluxes and the correlation of their associated genes within the co-expression network.
Subject to 
,
,
for all 
,
for all ,where the matrix 
includes submatrices 
� and 
�, which correspond to the columns of matrix S that represent irreversible and reversible reaction fluxes, respectively. The vector 
contains components of irreversibly and reversibly oriented fluxes, where 
� represents the reversible component. The function 
converts a gene expression value 
� into a corresponding flux bound value.
includes submatrices � and �, which correspond to the columns of matrix S that represent irreversible and reversible reaction fluxes, respectively. The vector contains components of irreversibly and reversibly oriented fluxes, where � represents the reversible component. The function converts a gene expression value � into a corresponding flux bound value.The vector 
encompasses components of irreversible and reversible reaction fluxes, where 
and 
denote the irreversible and reversible fluxes, respectively. The vector 
is initialized as zeros, with a single one placed at the position of the reaction of interest (biomass flux). The symbol 
represents the potential maximum biomass predicted using the E-flux method [2]. The parameter 
(0,1] determines the proportion of biomass required to evaluate the organisms' vitality, and in this study, α is set to 1. The set 𝑅𝑒𝑣 consists of reaction pairs derived from the same reversible reaction flux. The term 𝑀𝑗 represents the maximum gene expression value for reaction flux 𝑗.
encompasses components of irreversible and reversible reaction fluxes, where and denote the irreversible and reversible fluxes, respectively. The vector is initialized as zeros, with a single one placed at the position of the reaction of interest (biomass flux). The symbol represents the potential maximum biomass predicted using the E-flux method [2]. The parameter (0,1] determines the proportion of biomass required to evaluate the organisms' vitality, and in this study, α is set to 1. The set 𝑅𝑒𝑣 consists of reaction pairs derived from the same reversible reaction flux. The term 𝑀𝑗 represents the maximum gene expression value for reaction flux 𝑗.In this model, 
and 
represent the transformed flux values for reactions i and j, respectively. The set 𝑅 includes reaction pairs with genes linked in the co-expression network. The objective function is a summation of the products, specifically for reactions i and j corresponding to genes connected within the co-expression network.
and represent the transformed flux values for reactions i and j, respectively. The set 𝑅 includes reaction pairs with genes linked in the co-expression network. The objective function is a summation of the products, specifically for reactions i and j corresponding to genes connected within the co-expression network. REQUIREMENT
- Matlab (version 2018a or better)
- Cobra Toolbox
- Gurobi solver (version 9.0.1 or better, free academic)
- Gene expression profile in .csv file
(Note that the first column of gene expression data should have gene symbols/names used in the GPR association of the genome scale metabolic model. First row of gene expression data should have condition names.)
initCobraToolbox(false) % false, as we don't want to update
Load the expression data that will be used for the simulation. For this tutorial, we have choosen to use E. coli Microarray-based gene expression data (downloaded from http://systemsbiology.ucsd.edu/InSilicoOrganisms/Ecoli/EcoliExpression2)
modelFileName = 'ecoli_core_model.mat';
modelDirectory = getDistributedModelFolder(modelFileName); %Look up the folder for the distributed Models.
modelFileName= [modelDirectory filesep modelFileName]; % Get the full path. Necessary to be sure, that the right model is loaded
model = readCbModel(modelFileName);
Load the expression data that will be used for the simulation. For this tutorial, we have chosen to use E. coli microarray-based gene expression data
fileGeneName = 'gene_exp.csv';
fileDir = fileparts(which(fileGeneName));
cd(fileDir);
[exp, genetxt] = xlsread([fileDir filesep fileGeneName]);
The integration of the co-expression network and metabolic model is completed using the function ICONGEMs. The inputs are: a loaded model file, an expression array (exp), genetxt, a row vector of conditions for calculating flux distribution (with the default set to all conditions), and a threshold for constructing the co-expression network (default value is 0.9). The alpha value represents the proportion of biomass (a value in the range (0,1], with the default set to 1).
solution = ICONGEMs(model, exp, genetxt);
Using optional inputs:
% set parameter
condition = 1:size(exp,2);
threashold = 0.9;
alpha = 0.99;
Call ICONGEMs function:
solution1 = ICONGEMs(model, exp, genetxt, condition, threshold,alpha);
After the algorithm is finished, the solution for the predicted metabolic fluxes will be added to the workspace. Numerical flux values can be examined in more detail by double-clicking the solution. Moreover, the output of this algorithm is reported in the result.csv file.
REFERENCES
1. Paklao, T., Suratanee, A. & Plaimas, K. ICON-GEMs: integration of co-expression network in genome-scale metabolic models, shedding light through systems biology. BMC Bioinformatics 24, 492 (2023). https://doi.org/10.1186/s12859-023-05599-0.
2. Colijn C, Brandes A, Zucker J, Lun DS, Weiner B, Farhat MR, Cheng T-Y, Moody DB, Murray M, Galagan JE. Interpreting expression data with metabolic flux models: predicting mycobacterium tuberculosis mycolic acid production. PLoS Comput Biol. 2009;5(8):e1000489.