Rfastcormics

GPRrulesMapper(exp, x)

Evaluate Gene-Protein-Reaction (GPR) rules by mapping gene expression values to reactions using logical operators

USAGE:

[res] = GPRrulesMapper (exp, x)

INPUTS:

• exp – cell array with the expression to evaluate from the rules field

• x – value to plug into the expression

OUTPUTS:

res – the numeric value resulting from evaluating the expression

constrainModelOnMedium(model, mediumMets, notMediumConstrained, biomassReaction, functionToKeep)

Constrain a metabolic model based on a defined medium by restricting the exchange reactions, while preserving essential functions

USAGE:

[model] = constrainModelOnMedium (model, mediumMets, notMediumConstrained, biomassReaction, functionToKeep)

INPUTS:

• model – COBRA model structure with the following fields:

  • .S - m x n stoichiometric matrix

  • .lb - n x 1 lower flux bounds

  • .ub - n x 1 upper flux bounds

  • .rxns - n x 1 cell array of reaction identifiers

• mediumMets – cell array of metabolites defining the growth medium

• notMediumConstrained – list of reactions that should not be constrained by the medium

• biomassReaction – reaction identifier of the biomass reaction to preserve

• functionToKeep – list of reactions that must remain active

OUTPUTS:

model – constrained metabolic model

createFitL(xi, rest)

Create a Gaussian fit of the residual (left) curve used for FPKM discretization

USAGE:

[fitresult, gof] = createFitL (xi, rest)

INPUTS:

• xi – x values (log2(FPKM) grid) for the fit

• rest – y values (residual density) to fit

OUTPUTS:

• fitresult – a cfit object representing the Gaussian fit

• gof – structure with goodness-of-fit information

See also: FIT, CFIT, SFIT

createFitR(x, hybrid_curve)

Create a Gaussian fit of the hybrid (right) curve used for FPKM discretization

USAGE:

[fitresult, gof] = createFitR (x, hybrid_curve)

INPUTS:

• x – x values (log2(FPKM) grid) for the fit

• hybrid_curve – y values (hybrid density curve) to fit

OUTPUTS:

• fitresult – a cfit object representing the Gaussian fit

• gof – structure with goodness-of-fit information

See also: FIT, CFIT, SFIT

discretizeFPKM(fpkm, colnames, figflag, pathFigures, fileFormat)

Discretize gene expression data (FPKM values) into three categories, expressed, not expressed, and unknown, based on a zFPKM transformation and half-Gaussian density fitting

USAGE:

[discretized, scaledExpression] = discretizeFPKM (fpkm, colnames, figflag, pathFigures, fileFormat)

INPUTS:

• fpkm – m x n matrix or table of FPKM expression values (genes x samples)

• colnames – 1 x n cell array of sample names

OPTIONAL INPUTS:

• figflag – flag to plot and save figures (default: 0)

• pathFigures – path to save the figures (default: current folder)

• fileFormat – file format used to store the figures (default: ‘.png’)

OUTPUTS:

• discretized – m x n matrix of discretized values:

  • 1 - expressed

  • 0 - unknown

  • -1 - not expressed

• scaledExpression – scaled data after the zFPKM transformation

findOrganicExRxns(model, biomassReaction, functionToKeep)

Identify the exchange reactions that are carbon sources, excluding the biomass reaction, the reactions supplying the medium, and the reactions listed in the functionToKeep input

USAGE:

[exOrgaRxns, ExOrgaInd] = findOrganicExRxns (model, biomassReaction, functionToKeep)

INPUTS:

• model – COBRA model structure with the following fields:

  • .S - m x n stoichiometric matrix

  • .rxns - n x 1 cell array of reaction identifiers

  • .mets - m x 1 cell array of metabolite identifiers

  • .metFormulas - m x 1 cell array of metabolite chemical formulas

• biomassReaction – reaction identifier of the biomass reaction to exclude from constraining

• functionToKeep – cell array of reaction identifiers to keep unconstrained

OUTPUTS:

• exOrgaRxns – cell array of the candidate exchange reaction identifiers (biomass and functionToKeep reactions excluded)

• ExOrgaInd – indices of the organic exchange reactions among the candidates

fixIrr(model)

Convert irreversible backward reactions into irreversible forward reactions

USAGE:

[model] = fixIrr (model)

INPUTS:

model – COBRA model structure with the following fields:

• .S - m x n stoichiometric matrix

• .lb - n x 1 lower flux bounds

• .ub - n x 1 upper flux bounds

• .rxns - n x 1 cell array of reaction abbreviations

OUTPUTS:

model – model with corrected reversibilities, with fields:

• .rev - n x 1 reaction reversibility indicator (1 reversible, 0 irreversible)

• .lb - updated n x 1 lower flux bounds

• .ub - updated n x 1 upper flux bounds

• .S - stoichiometric matrix with backward reactions flipped to forward

mapExpressionToModel(model, data, dico, rownames, processTranscripts)

Map the expression data to a model following the GPR rules (Pacheco et al., 2019)

USAGE:

[mapping] = mapExpressionToModel (model, data, dico, rownames, processTranscripts)

INPUTS:

• model – COBRA model structure with the following fields:

  • .rxns - n x 1 cell array of reaction abbreviations

  • .genes - cell array of gene identifiers

  • .rules - cell array of GPR rules in x(i) logical form

• data – expression or discretized values for the samples, size(data, 1) is the number of gene IDs and size(data, 2) is the number of samples

• dico – table with corresponding gene identifier information, used to map the gene IDs to the genes in the model; can contain multiple columns with different identifiers

• rownames – cell array with the gene IDs

OPTIONAL INPUTS:

processTranscripts – 0 for inactive (default), 1 for active - if active, consider gene names of the model without any numbers after a “.”

OUTPUTS:

mapping – matrix of the expression values mapped to the reactions according to the GPR rules, size(mapping, 1) equals the number of reactions and size(mapping, 2) equals size(data, 2)

rFastcormics(model, discretized, rownames, dico, biomassReactionName, consensusProportion, epsilon, optionalSettings, fillingMediumFlag, adaptiveScalingFlag)

Reconstruct a context-specific model (tissue, cell type, or any context) from RNAseq data and a generic reconstruction using the rFASTCORMICS algorithm (Pacheco et al. 2019)

USAGE:

[contextSpecificModel, retainedRxns, indicesCompletedCoreOrig] = rFastcormics (model, discretized, rownames, dico, biomassReactionName, consensusProportion, epsilon, optionalSettings, fillingMediumFlag, adaptiveScalingFlag)

REQUIREMENTS:

Statistics and Machine Learning Toolbox, and Curve Fitting Toolbox

INPUTS:

• model – COBRA model structure with the following fields:

  • .S - m x n stoichiometric matrix

  • .lb - n x 1 lower flux bounds

  • .ub - n x 1 upper flux bounds

  • .rxns - n x 1 cell array of reaction abbreviations

  • .rules - cell array of GPR rules in x(i) logical form

  • .subSystems - reaction subsystem assignments

  • .metFormulas - m x 1 cell array of metabolite chemical formulas

• discretized – discretized expression values, size(discretized, 1) is the number of genes and size(discretized, 2) is the number of samples

• rownames – cell array with the gene IDs

• dico – table with gene identifier information used to map the rownames to the model genes; can contain multiple columns with different identifiers:

  • .Properties - table metadata; the VariableNames identify the column that matches the gene IDs

• biomassReactionName – reaction identifier of the objective (biomass) reaction

OPTIONAL INPUTS:

• consensusProportion – rate of samples that must (not) express a gene for it to be considered (not) expressed in the context of interest (default 0.9)

• epsilon – smallest flux considered nonzero (default 1e-4)

• optionalSettings – structure with optional constraints:

  • .func - cell array of reaction abbreviations that should carry a flux

  • .medium - cell array of metabolite abbreviations defining the growth medium

  • .notMediumConstrained - reaction abbreviations not in the medium that must be retained

• fillingMediumFlag – fill the medium with supplementary reactions when the provided medium is not sufficient to fulfill the objective; 1 for active (default), 0 for inactive

• adaptiveScalingFlag – adaptive scaling of the flux values (see LP10); 0 for inactive (default), 1 for active

OUTPUTS:

• contextSpecificModel – context-specific model, reduced to the retained reactions and associated genes

• retainedRxns – indices of the retained reactions in the input model

• indicesCompletedCoreOrig – indices of the core reactions in the input model

Example

[contextSpecificModel, retainedRxns, indicesCompletedCoreOrig] = rFastcormics(model, discretized, rownames, dico, biomassReactionName)