rMTA¶

MTA(model, rxnFBS, Vref, varargin)¶

Calculate Metabolic Transformation Analysis (MTA) using the the solver CPLEX. Code was prepared to be able to be stopped and be launched again by using a temporary file called ‘temp_MTA.mat’. Outputs are cell array for each alpha (one simulation by alpha). It there is only one alpha, content of cell will be returned The code here has been based on:

Yizhak, K., Gabay, O., Cohen, H., & Ruppin, E. (2013). ‘Model-based identification of drug targets that revert disrupted metabolism and its application to ageing’. Nature communications, 4. http://www.nature.com/ncomms/2013/131024/ncomms3632/full/ncomms3632.html

USAGE:

[TSscore,deletedGenes,Vout] = MTA(model, rxnFBS, Vref, alpha, epsilon, varargin)

INPUTS:

model: Metabolic model structure (COBRA Toolbox format). rxnFBS: Array that contains the desired change: Forward,

Backward and Unchanged (+1;0;-1). This is calculated from the rules and differential expression analysis.

Vref: Reference flux of the source state.

OPTIONAL INPUTS:

alpha: Numeric value or array. Parameter of the quadratic

problem (default = 0.66)

epsilon: Numeric value or array. Minimun perturbation for each

reaction (default = 0)

rxnKO: Binary value. Calculate knock outs at reaction level

instead of gene level. (default = false)

listKO: Cell array containing the name of genes or

reactions to knockout. (default = all)

timelimit: Time limit for the calculation of each knockout.

(default = inf)

SeparateTranscript: Character used to separate different transcripts of a gene. (default = ‘’).

Examples:

SeparateTranscript = ‘’
- gene 10005.1 ==> gene 10005.1
- gene 10005.2 ==> gene 10005.2
- gene 10005.3 ==> gene 10005.3
SeparateTranscript = ‘.’
- gene 10005.1
- gene 10005.2 ==> gene 10005
- gene 10005.3

numWorkers: Integer: is the maximun number of workers

used by the solver. 0 = automatic, 1 = sequential, > 1 = parallel. (default = 0)

FORCE_CPLEX: 1 to force CPLEX solver, 0 (default) for COBRA

solver.

printLevel: Integer. 1 if the process is wanted to be shown

on the screen, 0 otherwise. (default = 1)

OUTPUTS:

TSscore: Transformation score by each transformation deletedGenes: The list of genes/reactions removed in each knock-out Vres: Matrix of resulting fluxes

MTA_MIQP(OptimizationModel, KOrxn, varargin)¶

Returns the OptimizationModel solution of a particular MTA problem and an specific model

USAGE:

[v_res, success, unsuccess] = MTA_MIQP (OptimizationModel, KOrxn, numWorkers, timeLimit, printLevel)

INPUT:

OptimizationModel: Cplex Model struct KOrxn: perturbation in the model (reactions) numWorkers: number of threads used by Cplex. FORCE_CPLEX: 1 to force CPLEX solver, 0 (default) for COBRA

solver.

printLevel: 1 if the process is wanted to be shown on the: screen, 0 otherwise. Default: 1.

OUTPUTS:

Vout: Solution flux of MIQP formulation for each case solution: Cplex solution struct

MTA_TS(Vout, Vref, rxnFBS)¶

Returns the TS score of a particular solution of the MTA perturbation algorithm.

USAGE:

score = MTA_TS(v_res,vref,Model,success,unsuccess)

INPUT:: Vout: Solution flux of MIQP formulation for each case Vref: Reference flux of source state rxnFBS: Array that contains the desired change: Forward,

Backward and Unchanged (+1;0;-1).
OUTPUTS:: score: TS score for each case

buildMTAproblemFromModel(model, rxnFBS, Vref, varargin)¶

Returns the COBRA Optimization model needed to perform the MTA

USAGE:

OptimizationModel = buildMTAproblemFromModel(model,rxnFBS,Vref,alpha,epsilon)

INPUT:

model: Metabolic model (COBRA format) rxnFBS: Forward, Backward and Unchanged (+1;0;-1) values

corresponding to each reaction.

Vref: Reference flux of the source state. alpha: parameter of the quadratic problem (default = 0.66) epsilon minimun disturbance for each reaction, (default = 0)

OUTPUTS:

OptimizationModel: COBRA model struct that includes the: stoichiometric contrains, the thermodinamic constrains and the binary variables.

calculateEPSILON(samples, rxnFBS, varargin)¶

Calculate the minimum required flux change (epsilon) for reactions to be considered as significantly changed with p-value of 0.05 The code below is based on the method presented in:

Yizhak, K., Gabay, O., Cohen, H., & Ruppin, E. (2013). Model-based identification of drug targets that revert disrupted metabolism and its application to ageing. Nature communications, 4, 2632.

USAGE:

epsilon = calculateEPSILON(samples, rxnFBS, ‘unique_epsilon’, false, ‘minimum’, 1e-3)

INPUT:

samples: Matrix of sampled fluxes. rxnFBS: Array that contains the desired change: Forward,

Backward and Unchanged (+1;0;-1). This is calculated from the rules and differential expression analysis.

OPTIONAL INPUTS:

varargin: ParameterName value pairs with the following options:

unique_epsilon: True = unique epsilon, False = each reaction has different epsilon (default=false)

minimum: Minimun value for epsilon requiered (default=1e-3)

OUTPUT:

epsilon: Numeric value or array with the epsilon for the different reactions

calculateGeneKOMatrix(model, varargin)¶

Build a rxn-gene matrix such that the i-th column indicates what reactions become inactive because of the i-th gene’s knock-out.

USAGE:

geneKO = calculateGeneKOMatrix(model, SeparateTranscript, printLevel)

INPUT:

model: The COBRA Model structure

OPTIONAL INPUTS:

SeparateTranscript Character used to separate

different transcripts of a gene. (default = ‘’) Examples:

SeparateTranscript = ‘’

gene 10005.1 ==> gene 10005.1

gene 10005.2 ==> gene 10005.2

gene 10005.3 ==> gene 10005.3

SeparateTranscript = ‘.’

gene 10005.1

gene 10005.2 ==> gene 10005

gene 10005.3

printLevel: Integer. 1 if the process is wanted to be shown

on the screen, 0 otherwise. (default = 1)

OUTPUT:

geneKO: Struct which contains matrix with blocked: reactions for each gene in the metabolic model, name of reactions and name of genes.

diffexprs2rxnFBS(model, diff_exprs, Vref, varargin)¶

Returns Forward - Backward - Unchanged (+1;0;-1) for each reaction. (+1)R_f (-1)R_b (0)unchanged

USAGE:

rxnFBS = diffexprs2rxnFBS(model, diff_exprs, Vref, ‘SeparateTranscript’, ‘’, ‘logFC’, 0, ‘pval’,0.05)

INPUT:

model: The COBRA Model structure diff_exprs: MATLAB Table including the information of the

differentially expressed genes. Required columns (with theses names):

gene ( ID of gene, same as in the meabolic model)

logFC (SOURCE VS TARGET)

pval (p-value or adjusted-p-value)

Vref: Reference flux of the model

OPTIONAL INPUTS:

varargin: ParameterName value pairs with the following options:

SeparateTranscript: Character used to separate different transcripts of a gene. (default: ‘’)

logFC: minimum log2 (fold change) requiered (default = 0)

pval: maximum p-value admited (default = 0.05)

OUTPUT:

rxnFBS: array containting the information of altered: reactions: Forward - Backward - Unchanged (+1;0;-1);

rMTA(model, rxnFBS, Vref, varargin)¶

Calculate robust Metabolic Transformation Analysis (rMTA) using the solver CPLEX. Code was prepared to be able to be stopped and be launched again by using a temporally file called ‘temp_rMTA.mat’. Outputs are cell array for each alpha (one simulation by alpha). It there is only one alpha, content of cell will be returned

USAGE:

[TSscore,deletedGenes,Vout] = rMTA(model, rxnFBS, Vref, alpha, epsilon, varargin)

INPUTS:

model: Metabolic model structure (COBRA Toolbox format). rxnFBS: Array that contains the desired change: Forward,

Backward and Unchanged (+1;0;-1). This is calculated from the rules and differential expression analysis.

Vref: Reference flux of the source state.

OPTIONAL INPUTS:

alpha: Numeric value or array. Parameter of the quadratic: problem (default = 0.66)
epsilon: Numeric value or array. Minimun perturbation for each: reaction (default = 0)
parameterK: Numeric value value. Parameter to calculate rTS: score. (default = 100)
rxnKO: Binary value. Calculate knock outs at reaction level: instead of gene level. (default = false)
listKO: Cell array containing the name of genes or: reactions to knockout. (default = all)

timelimit: Time limit for the calculation of each knockout. (default = inf) SeparateTranscript: Character used to separate

different transcripts of a gene. (default = ‘’) Examples:

SeparateTranscript = ‘’

gene 10005.1 ==> gene 10005.1

gene 10005.2 ==> gene 10005.2

gene 10005.3 ==> gene 10005.3

SeparateTranscript = ‘.’

gene 10005.1

gene 10005.2 ==> gene 10005

gene 10005.3

numWorkers: Integer: is the maximun number of workers: used by the solver. 0 = automatic, 1 = sequential, > 1 = parallel. (default = 0)
FORCE_CPLEX: 1 to force CPLEX solver, 0 (default) for COBRA: solver.
printLevel: Integer. 1 if the process is wanted to be shown: on the screen, 0 otherwise. (default = 1)

OUTPUTS:

TSscore: Transformation score by each transformation deletedGenes: The list of genes/reactions removed in each knock-out Vref: Matrix of resulting fluxes

rMTAsaveInExcel(xls_filename, TSscore, deletedGenes, alpha_values, varargin)¶

Save in Excel all the results calculated using rMTA.

USAGE:

rMTAsaveInExcel(output_file, TSscore, deletedGenes, alpha_values, varargin)

INPUTS:

xls_filename: name of the resulting excel file TSscore: Transformation score by each transformation deletedGenes: The list of genes/reactions removed in each knock-out alpha_values: Numeric value or array. Parameter of the quadratic

problem (default = alpha_1, alpha_2,…)

OPTIONAL INPUTS:

gene_info: Table with the gene names and additional: information. It needs a column called ‘gene’, which contains the same information as the deletedGenes variable.
differ_genes: Table with the gene differential expression: information. It needs a column called ‘gene’, which contains the same information as the deletedGenes variable.
RankingGeneID: Caracter array with the ID of the gene which score: is desired to know.