BaMFA: Bayesian Metabolic Flux Analysis

Bayesian Metabolic Flux Analysis MATLAB package by Markus Heinonen, Maria Osmala, Henrik Mannerström, Juho Rousu, Samuel Kaski and Harri Lähdesmäki from the CSB research group of Aalto University.

The corresponding paper: Bayesian Metabolic Flux Analysis reveals intracellular flux couplings (arXiv)

Constructs a Bayesian metabolic model and samples the flux posterior: all flux vectors that are compatible with steady-state, flux upper and lower bounds and flux measurements. The flux posterior is represented by a set of flux vectors that can be analysed for flux distributions or flux pair distributions. Uses Gibbs sampling to characterise the posterior.

Genome-scale models are supported, but sampling time can be several hours.

This is a Bayesian drop-in replacement to standard flux balance analysis (FBA), metabolic flux analysis (MFA) and flux variability analyses (FVA).

Flux distributions

Below is visualised nine example flux pair distributions in three conditions (red/blue/green). The black dot is the classic FBA solution, while the scatter plots indicate all possible flux states (from functions bfma and plotfluxpair2, See demo.m).

Package

Requirements:

• MATLAB
• OpenCobra MATLAB toolbox

Main functions:

• sol = bfba(model,fba)
  • runs Bayesian FBA and returns set of optimal growth flux vectors
• sol = bmfa(model,meas,errs)
  • runs Bayesian MFA and returns sets of flux vectors compatible with measurements meas with errors errs
• plotfluxes(model,sol,fluxinds)
  • plots univariate flux distributions of fluxes fluxinds
• plotfluxpair2(model,sol,indpairs)
  • plots bivariate flux pair distributions (see also plotfluxpair)

Tutorial

Run demo.m in MATLAB or run

% run FBA on the ecoli core metabolism (77 reactions)

addpath code;
initCobraToolbox;

% read model
model = readCbModel('models/Ec_core_flux1','fileType','SBML');

% run baseline FBA
fba = optimizeCbModel(model);

% sample, uses by default 10 chains of 200 samples with thinning = 50
sol = bfba(model, fba);

%% visualise

% plot first 15 flux distributions with fba
figure; plotfluxes(model, sol, 1:15, fba);

% plot 7x7 flux pair grid
figure; plotfluxpair(model, sol, [10 2 40 42 45 53 55], fba);

% plot 8 example flux pairs
figure; plotfluxpair2(model, sol, [30 40; 30 41; 40 47; 15 17; 4 5; 1 6; 10 18; 42 45], fba);