Moomin

moomin(model, expression, varargin)[source]

MOOMIN (Pusa et al., 2019) generates a hypothesis of a metabolic shift using a metabolic network and differential expression data. Based on changes in the expression of genes, each reaction is given a colour to indicate an increase, a decrease, or no change in flux.

USAGE

[model, MILPsolutions, MILPproblem] = moomin (model, expression, varargin)

INPUTS

  • model – input model (COBRA model structure)

  • expression – structure with the following fields

    • .GeneID - vector of gene IDs

    • .PPDE - vector of posterior probabilities of differential

    expression * .FC - vector of log fold changes

Optional parameters can be entered the standard MATLAB way with parameter name followed by parameter value: i.e. ,’pThresh’, 0.9)

OPTIONAL INPUTS

  • pThresh – threshold for differential expression (default 0.9)

  • alpha – alpha parameter of the weight function, a higher value means less evidence is needed for a change to be inferred (default 3) 0 < alpha < 10, a good range is [0.5, 3]

  • stoichiometry – Boolean to choose if stoichiometry is considered (default 1)

  • enumerate – integer to determine the maximum number of alternative optimal solutions that are enumerated (default 1 ie only one solution)

  • precision – integer to determine up to how many significant numbers the weights are evaluated. Will influence the uniqueness of weight values. (default 7)

  • solverTimeLimit – time limit for the MILP solver (default 1000)

  • solverPrintLevel – print level parameter for the MILP solver (default 0)

  • solverParameters – struct containing solver parameters to be passed on to solverCobraMILP

OUTPUTS

  • model – input model with additional fields containing outputs of the algorithm. Rows correspond to reactions. The colours are coded by

    2 - reverse red (r.red) ie increase, reaction in reverse 1 - red ie increase 0 - grey ie no change -1 - blue ie decrease -2 - reverse blue (r.blue) ie decrease, reaction in reverse 6 - yellow ie unspecified change in a reversible reaction

    Additional fields are

    • .inputColours - colours inferred solely based on data

    • .outputColours - matrix, colours inferred by the algorithm columns correspond to alternative optimal solutions

    • .weights - the reaction weights used in the algorithm

    • .leadingGenes - the genes “responsible” for the colour and weight of a reaction

    • .frequency - how often a reaction is coloured in a solution

    • .combined - an attempted consensus between all optimal solutions (colour differs between solutions -> 6)

    • .PPDE - PPDEs of the model genes. -1 for missing values

    • .FC - fold changes of the model genes. 0 for missing values

  • MILPsolutions raw outputs of ‘solveCobraMILP’

  • MILPproblem the final MILP-problem solved (or was attempted to be solved)

Pusa et al. (2019). MOOMIN – Mathematical explOration of ‘Omics data on a MetabolIc Network.

weightFunction(PPDE, alpha, beta, pThresh)[source]

Gives the gene/reaction weight used in the MOOMIN-algorithm.

USAGE

weight = weightFunction (PPDE, alpha, beta, pThresh)

INPUTS

  • PPDE – posterior probability of differential expression

  • alpha – parameter to control the relationship between the positive and negative weights, and the sparseness of the inferred solutions

  • beta – shape parameter for the weight function

  • pThresh – threshold for differential expression

OUTPUT

weight – the value of the weight function given the inputs

writeMoominOutput(model, fileName, varargin)[source]

Writes the output of MOOMIN into a text file.

USAGE

writeMoominOutput (model, fileName, varargin)

INPUTS

  • model – COBRA model structure output by MOOMIN

  • filename – name of the output file

Optional parameters can be entered the standard MATLAB way with parameter name followed by parameter value: i.e. ,’nSolution’, 3)

OPTIONAL INPUTS

nSolution: number of the solution that is output (default 1) format: control the output format

  • ‘standard’: default output, a tab-delimited file giving output

    colours for reactions for one solution

  • ‘full’: a tab-delimited file listing one solution as well as other

    information

  • ‘json’: a .json file of colours for one solution

type: choose which kind of solution is written

  • ‘output’: one solution, as specified by nSolution (default)

  • ‘input’: input ie a priori colours

  • ‘combined’: consensus (combination of all optimal solutions found)

  • ‘frequency’: how often a reaction is coloured

string Boolean to control if colours are written out or coded by numbers (default 1)