Sven Thiele
Meneco is a tool for metabolic network completion. Within a qualitative approach that describes the bio-synthetic capacities of metabolic networks, Meneco uses qualitative constraints to express the producibility for a set of metabolites. Meneco can be used to check whether a network provides the synthesis routes to comply with the required functionality described by the producibility constraints. In particular it tests whether it is possible to synthesize so called target metabolites from a set of seed metabolites.
For networks that fail this test Meneco can attempt to complete the network by importing reactions from a metabolic reference network such that the resulting network provides the required functionality. Meneco can identify unproducible target metabolites and computes minimal extensions to the network that satisfy the producibility constraints. Additionally, it can compute the union and intersection of all minimal networks extensions without enumerating all minimal network extensions.
Meneco is a Python application that uses the power of answer set solving technology to compute its solution.
Therefore it depends on the Clyngor library, a python wrapper for the solvers from the Potassco Answer Set Solving Collection.
All dependencies are automatically resolved and installed via pip, the recommended package installer for the Python Package index where the software is hosted.
Meneco runs on the Linux and Mac OS operating systems Windows is currently not supported.
pipYou can install the meneco package by running:
pip install --user meneco
On Linux the executable scripts can then be found in ~/.local/bin
and on Mac OS the scripts are under /Users/YOURUSERNAME/Library/Python/3.6/bin.
Typical usage is:
meneco -d draftnetwork.sbml -s seeds.sbml -t targets.sbml -r repairnetwork.sbml
For more options you can ask for help as follows:
usage: meneco [-h] -d DRAFTNET -s SEEDS -t TARGETS [-r REPAIRNET]
[--enumerate] [--json]
optional arguments:
-h, --help show this help message and exit
-d DRAFTNET, --draftnet DRAFTNET
metabolic network in SBML format
-s SEEDS, --seeds SEEDS
seeds in SBML format
-t TARGETS, --targets TARGETS
targets in SBML format
-r REPAIRNET, --repairnet REPAIRNET
perform network completion using REPAIRNET a metabolic
network in SBML format
--enumerate enumerate all minimal completions
--json produce JSON output
You can use Meneco from python by calling the command run_meneco() with the paths of files as input arguments and a boolean value for the enumeration (True for the enumeration, else False) :
from meneco import run_meneco
result = run_meneco(draftnet="toy/draft.sbml",
seeds="toy/seeds.sbml",
targets="toy/targets.sbml",
repairnet="toy/repair.sbml",
enumeration=False,
json=True)
The output will be the set of unproducible targets, reconstructable targets, a dictionnary of essentials reactions for each target, one minimal solution, the set of reactions belonging to the intersection of solutions, the set of reactions belonging to the union of solutions and a list of lists corresponding to the reactions for each solution (if enumeration == True).
For a step by step demonstration on how to use Meneco as a library, have a look at our notebooks here or here.
Meneco works with two kinds of data. The first is metabolic reaction data representing metabolic reactions from the draft network and the reference database (i.e. the metacyc db). The second is the information about seed and target metabolites.
The metabolic reaction data must be presented in Systems Biology Markup Language format `SBML} as shown below.
Here is how one could represent a metabolic reaction network in SBML:
<?xml version="1.0" encoding="UTF-8"?>
<sbml level="2" version="3" xmlns="http://www.sbml.org/sbml/level2/version3">
<model name="EnzymaticReaction">
<listOfCompartments>
<compartment id="cytosol" size="1e-14"/>
</listOfCompartments>
<listOfSpecies>
<species compartment="cytosol" id="ES" initialAmount="0" name="ES"/>
<species compartment="cytosol" id="P" initialAmount="0" name="P"/>
<species compartment="cytosol" id="S" initialAmount="1e-20" name="S"/>
<species compartment="cytosol" id="E" initialAmount="5e-21" name="E"/>
</listOfSpecies>
<listOfReactions>
<reaction id="veq">
<listOfReactants>
<speciesReference species="E"/>
<speciesReference species="S"/>
</listOfReactants>
<listOfProducts>
<speciesReference species="ES"/>
</listOfProducts>
</reaction>
<reaction id="vcat" reversible="true">
<listOfReactants>
<speciesReference species="ES"/>
</listOfReactants>
<listOfProducts>
<speciesReference species="E"/>
<speciesReference species="P"/>
</listOfProducts>
</reaction>
</listOfReactions>
</model>
</sbml>
In this example, the model has the identifier EnzymaticReaction.
The model contains one compartment (with identifier cytosol),
four metabolites (with species identifiers ES, P, S, and E), and two reactions (veq and vcat).
The elements in the listOfReactants and listOfProducts in each reaction refer to the names of elements listed in the listOfSpecies.
The correspondences between the various elements is explicitly stated by the speciesReference elements.
The reaction vcat has the attribute reversible set to true.
Note that meneco will only treat a reaction as reversible if this attribute is set to true.
If the attribute is not set, the default assumption is that a reaction is irreversible.
Thus, the reaction veq is treated as irreversible.
The SBML file may contain additional informations (like initialAmount="1e-20") which will be ignored by Meneco.
Also information about seed and target metabolites can be specified using SBML files. For convenience reasons Meneco takes two files one which presents the specification of the seed metabolites and one file for the target metabolites. Here is how such a file could look like:
<?xml version="1.0" encoding="UTF-8"?>
<sbml level="2" version="3" xmlns="http://www.sbml.org/sbml/level2/version3">
<model id="enzreact_ntrexp_001" name="set of seed metabolites 1">
<listOfSpecies>
<species id="E" name="This crazy metabolite E"/>
<species id="S" name="This funky metabolite S"/>
</listOfSpecies>
</model>
</sbml>
In this example two metabolites (with identifiers E and S) are specified. It is important that these identifiers match with the identifiers in the description of the reaction network. Depending on whether this description is passed as 3rd or 4th argument Meneco regards the specified metabolites as seeds resp. targets.
If you do not set the flag --json the result is written in markdown like format to stdout.
The output of Meneco can be redirected into a file using the > operator.
For example to write the results shown below into the file result.md type:
meneco.py -d draftnet.sbml -s seeds.sbml -t targets.sbml -r repairdb.sbml --enumerate > result.md
While Meneco is running you get some status messages about what it does.
Reading draft network ...
Reading seeds ...
Reading targets ...
Checking draftnet for unproducible targets
Reading repair db ...
Checking draftnet + repairnet for unproducible targets
Computing essential reactions for M_cu2_c
Computing essential reactions for M_cl_c
Adding essential reactions to network
Computing one minimal completion to produce all targets
Computing common reactions in all completion with size 3
Computing union of reactions from all completion with size 3
Computing all completions with size 3
In the following we will dissect the output generated by Meneco.
Draft network file: draftnet.sbml
Seeds file: seeds.sbml
Targets file: targets.sbml
The first 3 lines document which files have been used as the metabolic reaction network and to specify the seeds and targets.
7 unproducible targets:
M_so4_c
M_cl_c
M_cu2_c
M_pe161_c
M_thf_c
M_ribflv_c
M_utp_c
In this section Meneco outputs which target metabolite are producible given the seeds and the reactions of the draft network. In this case the result shows 7 target metabolites which are not producible.
Given that unproducible target metabolites exist the next sections documents which of the targets remain unproducible even with the reactions from the repair database repairdb.sbml
and for which targets the metabolic pathways can be reconstructed.
Repair db file: repairdb.sbml
Still 5 unreconstructable targets:
M_so4_c
M_pe161_c
M_utp_c
M_ribflv_c
M_thf_c
2 reconstructable targets:
M_cu2_c
M_cl_c
In this example 5 metabolites remain unproducible, but for 2 targets Meneco is able to repair the synthesis pathways.
As a precomputation step Meneco reconstruct the production pathways for each single reconstructable target metabolite, and computing the essential reactions. Essential reaction in this context are reactions which always must be added to allow a synthesis of the target from the seeds. This intermediate result is reused in the subsequent computations as a solution that restores all targets must contain all reaction that are essential for each single target.
1 essential reactions for target M_cu2_c:
R_CU2tpp
1 essential reactions for target M_cl_c:
R_CLt3_2pp
Overall 2 essential reactions found:
R_CU2tpp
R_CLt3_2pp
In this example for each single target metabolite exist one essential reaction.
The section shows one minimal completion to produce all targets.
One minimal completion of size 3:
R_O2Stex
R_CU2tpp
R_CLt3_2pp
In this example the first minimal completion contains 3 reactions.
Given the size of a minimal completion.s. Meneco outputs the intersection and union of all minimal completion
Intersection of cardinality minimal completions:
R_CU2tpp
R_CLt3_2pp
Union of cardinality minimal completions:
R_O2tex
R_O2Stex
R_CU2tpp
R_CLt3_2pp
In our case we get 2 reactions in the intersection and 4 reactions in the union.
Finally, if the command line option --enumerate has been given,
all minimal completions are enumerated.
Completion 1:
R_O2Stex
R_CU2tpp
R_CLt3_2pp
Completion 2:
R_O2tex
R_CU2tpp
R_CLt3_2pp
Fortunately, our example has only two solutions. In general the number of solutions can be very high. Therfore, enumeration is only done if explicitly requested via command line option.
Meneco, a Topology-Based Gap-Filling Tool Applicable to Degraded Genome-Wide Metabolic Networks. (2017). PLOS Computational Biology. DOI
The genome-scale metabolic network of Ectocarpus siliculosus (EctoGEM): a resource to study brown algal physiology and beyond. (2014). The Plant Journal. DOI
Extending the Metabolic Network of Ectocarpus Siliculosus using Answer Set Programming. (2013). 12th International Conference on Logic Programming and Nonmonotonic Reasoning. DOI
Metabolic Network Expansion with Answer Set Programming. (2009). 25th International Conference on Logic Programming. DOI