Constraint-Based Reconstruction and EXascale Analysis

RepositoryTestsCoverageHow to contribute?

COBREXA is a toolkit for working with large constraint-based metabolic models, and running very large numbers of analysis tasks on these models in parallel. Its main purpose is to make the methods of Constraint-based Reconstruction and Analysis (COBRA) scale to problem sizes that require the use of huge computer clusters and HPC environments, which allows them to be realistically applied to pre-exascale-sized models.

In this package, you will find the usual COBRA-like functions that interface to underlying linear programming solvers. We use JuMP.jl as the unified interface for many solvers; you can plug in whichever compatible solver you want, including the popular Tulip.jl, GLPK.jl, OSQP.jl, and Gurobi.jl.

Development history of COBREXA.jl.

Quick start guide

You can install COBREXA from Julia repositories. Start julia, press ] to switch to the Packaging environment, and type:

add COBREXA

You also need to install your favorite solver supported by JuMP.jl, typing e.g.:

add Tulip

Alternatively, you may use prebuilt Docker and Apptainer images.

When the packages are installed, switch back to the "normal" julia shell by pressing Backspace (the prompt should change color back to green). After that, you can download a SBML model from the internet and perform a flux balance analysis as follows:

using COBREXA   # loads the package
using Tulip     # loads the optimization solver

# open the SBML file and load the contents

# run a FBA
fluxes = flux_balance_analysis_dict(model, Tulip.Optimizer)

The variable fluxes will now contain a dictionary of the computed optimal flux of each reaction in the model:

Dict{String,Float64} with 95 entries:
"R_EX_fum_e"    => 0.0
"R_ACONTb"      => 6.00725
"R_TPI"         => 7.47738
"R_SUCOAS"      => -5.06438
"R_GLNS"        => 0.223462
"R_EX_pi_e"     => -3.2149
"R_PPC"         => 2.50431
"R_O2t"         => 21.7995
"R_G6PDH2r"     => 4.95999
"R_TALA"        => 1.49698
⋮               => ⋮

Model variant processing

The main feature of COBREXA.jl is the ability to easily specify and process many analyses in parallel. To demonstrate, let's see how the organism would perform if some reactions were disabled independently:

# convert to a model type that is efficient to modify
m = convert(StandardModel, model)

# find the model objective value if oxygen or carbon dioxide transports are disabled
screen(m, # the base model
variants=[ # this specifies how to generate the desired model variants
[], # one with no modifications, i.e. the base case
[with_changed_bound("R_O2t", lower=0.0, upper=0.0)], # disable oxygen
[with_changed_bound("R_CO2t", lower=0.0, upper=0.0)], # disable CO2
[with_changed_bound("R_O2t", lower=0.0, upper=0.0),
with_changed_bound("R_CO2t", lower=0.0, upper=0.0)], # disable both
],
# this specifies what to do with the model variants (received as the argument x)
analysis = x ->
flux_balance_analysis_dict(x, Tulip.Optimizer)["R_BIOMASS_Ecoli_core_w_GAM"],
)

You should receive a result showing that missing oxygen transport makes the biomass production much harder:

4-element Vector{Float64}:
0.8739215022674809
0.21166294973372796
0.46166961413944896
0.21114065173865457

Most importantly, such analyses can be easily specified by automatically generating long lists of modifications to be applied to the model, and parallelized.

Knocking out each reaction in the model is efficiently accomplished:

# load the task distribution package, add several worker nodes, and load
# COBREXA and the solver on the nodes
using Distributed
@everywhere using COBREXA, Tulip

# get a list of the workers
worker_list = workers()

# run the processing in parallel for many model variants
res = screen(m,
variants=[
# create one variant for each reaction in the model, with that reaction knocked out
[with_changed_bound(reaction_id, lower=0.0, upper=0.0)]
for reaction_id in reactions(m)
],
analysis = model -> begin
# we need to check if the optimizer even found a feasible solution,
# which may not be the case if we knock out important reactions
sol = flux_balance_analysis_dict(model, Tulip.Optimizer)
isnothing(sol) ? nothing : sol["BIOMASS_Ecoli_core_w_GAM"]
end,
# run the screening in parallel on all workers in the list
workers = worker_list,
)

In result, you should get a long list of the biomass production for each reaction knockout. Let's decorate it with reaction names:

Dict(reactions(m) .=> res)

...which should output an easily accessible dictionary with all the objective values named, giving a quick overview of which reactions are critical for the model organism to create biomass:

Dict{String, Union{Nothing, Float64}} with 95 entries:
"ACALD"       => 0.873922
"PTAr"        => 0.873922
"ALCD2x"      => 0.873922
"PDH"         => 0.796696
"PYK"         => 0.864926
"CO2t"        => 0.46167
"EX_nh4_e"    => 1.44677e-15
"MALt2_2"     => 0.873922
"CS"          => 2.44779e-14
"PGM"         => 1.04221e-15
"TKT1"        => 0.864759
⋮             => ⋮

Basic and quick-start tutorials

Detailed tutorial content is available here.

Detailed listing of advanced tutorials is available here.

Example notebooks and workflows

Detailed notebook content is available here.

Contribution guide

If you wish to contribute code, patches or improvements to COBREXA.jl, please read the basic contribution guidelines and hints..

Acknowledgements

COBREXA.jl is developed at the Luxembourg Centre for Systems Biomedicine of the University of Luxembourg (uni.lu/lcsb), cooperating with the Institute for Quantitative and Theoretical Biology at the Heinrich Heine University in Düsseldorf (qtb.hhu.de).

The development was supported by European Union's Horizon 2020 Programme under PerMedCoE project (permedcoe.eu) agreement no. 951773.