Protocol: Pipeline for predicting therapeutic miRNA based on metabolic network
1. Pipeline using GIMME
2. Pipeline using our novel approach to reconstruct context specific network
The context dependent model to start from (e.g. a tissue specific cancer model), the expressionData for the condition, the miRNA_target matrix, and the two variables describing gene-reaction relationships (see the Protocol of reconstructing context dependent metabolic network)
For further analysis, the pipeline can also be applied to identify essential metabolic enzymes for a given system. Similar to application in miRNA:
Created by Ming Wu, 07/30/2012
I suggest reading Protocol [Novel approach to reconstruct context dependent metabolic network] before this protocol, because the pipeline is built based on the novel approach that I developed.
0. the miRNA-target relationship
A matrix is needed to provide: each row is a miRNA, for each miRNA, its targets (Entrez gene ID) in the metabolic system are listed as a vector.
An example variable is shown in here:
1. Pipeline using GIMME
Refer to our paper and protocol for reconstruct context dependent network to know GIMME.
our code for the pipeline is here:
- TissueS_miRNA_biomass.m
[MetaGeneTable,F0,vF1,diFF]=TissueS_miRNA_biomass(expressionData,miRNA_target, model);
Inputs:
gene expression data (MATLAB structure variable), miRNA_target matrix, and the context dependent model to start from (e.g. a tissue specific cancer model).
Results:
MetaGeneTable extracts metabolic gene targets from miRNA_target matrix.
F0 is the max value of objective function for the unperturbed model.
vF1 is a vector in which each element is the max value of objective function for the corresponding miRNA (overexpression/delivery of the miRNA)
diFF is a vector compare differences between perturbed and unperturbed model for each miRNA
2. Pipeline using our novel approach to reconstruct context specific network
Refer to our paper to know the theory of the approach.
our code for the pipeline is here:
- predictMIRNA_FC_single_flux.m
- UB_update_withFC.m
[diFF,flux00,fluxAll,diffF]=predictMIRNA_FC_single_flux(Model,expressionData, miRNA_target, ParsedGPR, Prxn);
The Inputs:
Results:
diFF is the reduction of maxachievablevalue of the objective function, e.g. the reduction of biomass production in cancer model for each miRNA.
flux00 is a vector representing the flux distribution for each reactions for the unperturbed model.
fluxAll are vectors representing flux distributions for every of the perturbed model corresponding to each miRNA. Each column is a flux vector for a miRNA perturbation.
diffF is the differences between each vector in fluxAll and flux00.
For further analysis, the pipeline can also be applied to identify essential metabolic enzymes for a given system. Similar to application in miRNA:
[diFF_metaEnzymes]=predictMIRNA_FC_single_flux(Model,expressionData, List_of_metaEnzyme, ParsedGPR, Prxn);
The only differences is the miRNA-target matrix is changed to a vector list of metabolic enzymes. This pipeline then can identify the essential metabolic enzyme whose downregulation will reduce biomass production rate.