Rumba

addMissingReactions(SampledModel, completeModel)[source]

Verify the consistency between the sampled model and the complete reference model by checking that all reactions from the complete model are in the sampled model, then copy the sampled flux points onto the complete model reaction set (points of missing reactions are left as zeros)

USAGE:

model = addMissingReactions (SampledModel, completeModel)

INPUTS:

  • SampledModel – Sampled model, with fields:

    • .rxns - reaction identifiers of the sampled model

    • .points - nRxns x nPoints matrix of sampled flux values

  • completeModel – The complete reference model, with fields:

    • .rxns - reaction identifiers of the complete model

OUTPUTS:

model – Consistent sampled model wrt the complete model, with fields:

  • .points - nRxns x nPoints sampled flux values mapped onto the complete model reactions (missing reactions zero)

classifyRxns(completeModel, sampledModel, maxMetConn, maxNumPnts, verboseTag, LoopRxnsToIgnore)[source]

Score, for each metabolite of the sampledModel, the incoming and outgoing reactions by the fraction of all incoming/outgoing flux that passes through the metabolite across the sampled flux points

USAGE:

[MetConnectivity, ConnectedMet] = classifyRxns (completeModel, sampledModel, maxMetConn, maxNumPnts, verboseTag, LoopRxnsToIgnore)

INPUTS:

  • completeModel – The complete reference model

  • sampledModel – Sampled model, with fields:

    • .points - nRxns x nPoints matrix of sampled flux values

  • maxMetConn – The maximum connectivity of a metabolite to consider. All branch points with a higher connectivity will be ignored (default = 30)

  • maxNumPnts – Maximum number of points to use from the sampled model. Extra points are removed to improve memory usage and speed up calculations (default = minimum number of points in the model or 500 points, whichever is smaller)

  • verboseTag – 1 = print out progress and use waitbars, 0 = print only minimal progress to screen

  • LoopRxnsToIgnore – List of rxns associated with loops within the model (default = reaction loops defined using FVA)

OUTPUTS:

  • MetConnectivity – A structure that presents, for each metabolite of the sampledModel, the following sets of fields:

    • ConnRxns - the reactions connected to the metabolite

    • Sij - stoichiometric coefficient for the metabolite in each reaction in ConnRxns

    • RxnScore - score for each reaction in ConnRxns

    • Direction - direction of reaction flux for each sample point

    • MetNotUsed - whether or not the metabolite is used in the condition

  • ConnectedMet – List of metabolites described in MetConnectivity

compareConditions(MetConnectivity1, ConnectedMet1, MetConnectivity2, ConnectedMet2)[source]

This function compare reaction score of both sampling condition and identify metabolites that change significantly

USAGE:

[MetsAndRxns, pVal_up, pVal_down, Dir_1, Dir_2] = compareConditions (MetConnectivity1, ConnectedMet1, MetConnectivity2, ConnectedMet2)

INPUTS:

  • MetConnectivity1 – A structure that present, for each metabolite present in the model sampled under first condition the following sets of fields:

    • ConnRxns - the reactions that are connected to the metabolite

    • Sij - The stoichiometric coefficient for the metabolite in each reaction in ConnRxns

    • RxnScore - Score for each reaction in ConnRxns

    • Direction - The direction of reaction flux for each sample point

    • MetNotUsed - Whether or not the metabolite is used in the condition

  • ConnectedMet1 – List of metabolites described in MetConnectivity1

  • MetConnectivity2 – Same as MetConnectivity1 but for model sampled under the second condition

  • ConnectedMet2 – List of metabolites described in MetConnectivity2

OUTPUTS:

  • MetsAndRxns – Cell arrays containing in the first column the list of metabolites that significantly change under both sampling conditions and in the second column the reactions that are connected to these metabolites

  • pVal_up – p-value associated to upregulated MetsAndRxns

  • pVal_down – p-value associated to downregulated MetsAndRxns

  • Dir_1 – reaction directionality for model sampled under first condition (1 producing metabolite, -1 consuming the metabolite)

  • Dir_2 – Same as Dir_1 but for model sampled under the second condition

normalizePoints(model1, model2, NormalizePointsParam, LoopRxnsToIgnore)[source]

Normalize the sampled points by the net network flux (NormalizePointsParam = 1) or by growth rate (NormalizePointsParam = 2)

USAGE:

[model1, model2] = normalizePoints (model1, model2, NormalizePointsParam, LoopRxnsToIgnore)

INPUTS:

  • model1 – Model sampled under the first condition, with fields:

    • .rxns - reaction identifiers

    • .rxnGeneMat - nRxns x nGenes reaction-to-gene mapping (rebuilt if absent)

    • .points - nRxns x nPoints matrix of sampled flux values

  • model2 – Model sampled under the second condition, with fields:

    • .rxns - reaction identifiers

    • .rxnGeneMat - nRxns x nGenes reaction-to-gene mapping (rebuilt if absent)

    • .points - nRxns x nPoints matrix of sampled flux values (scaled by the model1/model2 ratio)

    • .ub - nRxns x 1 upper flux bounds (scaled)

    • .lb - nRxns x 1 lower flux bounds (scaled)

  • NormalizePointsParam – Option to normalize sample points to (1) the same median of magnitude of flux through all non-loop gene-associated reactions, or (2) the optimal growth rate (default = 1)

  • LoopRxnsToIgnore – List of rxns associated with loops within the model (default = reaction loops defined using FVA)

OUTPUTS:

  • model1 – Normalized sampled model under first condition

  • model2 – Normalized sampled model under second condition

pValDistForModelOverlap(model1, model2)[source]

Compute for each reaction in common in both sampled modelw the magnitude of median flux value change

USAGE:

[rxnsInCommon, MedianChange] = pValDistForModelOverlap (model1, model2)

INPUTS:

  • model1 – Model sampled under the first condition, with fields:

    • .rxns - reaction identifiers

    • .points - nRxns x nPoints matrix of sampled flux values

  • model2 – Model sampled under the second condition, with fields:

    • .rxns - reaction identifiers

    • .points - nRxns x nPoints matrix of sampled flux values

OUTPUTS:

  • rxnsInCommon – Reactions shared by both sampled models

  • MedianChange – Magnitude of median flux value change for each reaction listed in ‘rxnsInCommon’

rumba(model1, model2, completeModel, sampling, maxMetConn, RxnsOfInterest, GenesOfInterest, NormalizePointsOption, PValCuttoff, MaxNumPoints, LoopRxnsToIgnore, verboseTag)[source]

RUMBA predicts which reactions significantly change their flux at metabolic branch points under two conditions

USAGE:

[RUMBA_outputs, UpRegulated, DownRegulated, MetConnectivity1, MetConnectivity2] = rumba (model1, model2, completeModel)

INPUTS:

  • model1 – Model under first condition, exchange reactions are constrained with the data related to the first condition. If model already sampled (‘sampling’ = 0), the sampling points are in an m x n matrix with m reactions and n points included as the .points field. Set ‘sampling’ = 1 to constrain and sample the model under the first condition. Fields used:

    • .points - nRxns x nPoints matrix of sampled flux values

    • .genes - gene identifiers

  • model2 – Model under second condition. Same format as model1, with the same .points and .genes fields.

  • completeModel – The complete reference model, used to verify consistency between the sampled models. Fields used:

    • .genes - gene identifiers

OPTIONAL INPUTS:

  • sampling – 0, if no sampling needed (default) 1, if sampling of the models under both conditions

  • maxMetConn – The maximum connectivity of a metabolite to consider. All branch points with a higher connectivity will be ignored (default = 30)

  • RxnsOfInterest – Reactions for which predictions are desired. Specifying only desired reactions speeds up algorithm (default = all reactions)

  • GenesOfInterest – Genes associated with the reactions of interest (default = all genes)

  • NormalizePointsOption – Option to normalize sample points to (1) the same median of magnitude of flux through all non-loop gene-associated reactions, or (2) the optimal growth rate. (default = 1)

  • PValCuttoff – P-value cutoff used to decide which changes in branch point flux to call significant (two- tailed p-value, so .05 will mean 0.25 on both tails). (default = 0.05)

  • MaxNumPoints – Maximum number of points to use form the sampled models. Extra points will be removed to improve memory usage and speed up calculations. (default = minimum number of points in the model or 500 points, whichever is smaller)

  • verboseTag – 1 = print out progress and use waitbars (default). 0 = print only minimal progress to screen.

  • LoopRxnsToIgnore – list of rxns associated with loop within the model, (default - reaction loops defined using FVA)

OUTPUTS:

  • RUMBA_outputs – Structure containing all information about each gene-reaction pair. For gene-reaction pair for which the differential branch-point information is possible to calculate: the list of connected metabolites (‘ConnectedMets’), the associated up-regulation p-value (‘pValue_up’), the associated down-regulation p-value (‘pValue_down’)and the reaction directionality for both model (‘direction’).The structure also contains the list of gene-reaction pairs for which no differential branch-point information can be determined (because of loops, unused pathways, etc.).

  • UpRegulated – first column : Gene-reaction pairs towards which flux is significantly upregulated during the shift; second column : list of metabolites connected to the gene-reaction pair; third column : Magnitude of absolutes flux change

  • DownRegulated – same structure as UpRegulated but for gene-reaction pairs from which flux is significantly downregulated during the shift

  • MetConnectivity1 – A structure that present for each metabolite present in model1 the following sets of fields:

    • ConnRxns - the reactions that are connected to the metabolite

    • Sij - The stoichiometric coefficient for the metabolite in each reaction in ConnRxns

    • RxnScore - Score for each reaction in ConnRxns

    • Direction - The direction of reaction flux for each sample point

    • MetNotUsed - Whether or not the metabolite is used in the condition

  • MetConnectivity2 – Same as MetConnectivity1 but for model2