Computing Elementary Flux Modes (EFM) with Answer Set Programming (ASP).

Elementary Flux Modes are minimal sets of enzymes that operate at steady state with all irreversible reactions proceeding in the appropriate direction.

The enumeration of EFMs is a difficult task. It requires the resolution of combinatorial problems on metabolic networks, and the integration of appropriate biological constraints to help calculations.

We propose to use the SAT-based power of ASP constraint logic programming resolution to reduce the hurdle of obtaining pathways of interest with EFMs on large-scale networks.

This tool uses clingo[LP] (clingoLP.py) and the Answer Set Programming (ASP) rules specified in solveLP.lp4 to compute the Elementary Flux Modes (EFMs) of a given metabolic network.

The input metabolic network should be in ASP format.

The module mparser should be used for converting metabolic networks into ASP rules.

Example executions of the tool are found in the bash file ./launch.sh, ./ext_launch.sh, ./mcs_launch.sh.

This module requires python version 3.

clingo version 5.4.0, compilation of clingo with Python is mandatory

conda install -c potassco clingo

conda install -c conda-forge future

pandas, numpy, scipy, sympy modules

conda install pandas
conda install numpy
conda install scipy
conda install sympy

clingo[LP] is a linear programming extension for the solver clingo, compatible with the IBM cplex solver.

If you don't have access to a cplex license, you may install the anaconda distribution :

conda install -c ibmdecisionoptimization cplex

Then, this extension only requires to use the script clingoLP.py, which is provided on this repository.

Here is how to use aspefm from a metabolic network file encoded in ASP :

clingo clingoLP.py [FILE] solve[LP].lp4 -c nstrict=0 -n 0 -c accuracy=10

Minimization heuristics are working with clingo extensions clingo[LP]:

To use them, add the following options when running clingo :

--heuristic Domain --enum-mode domRec

One can constrain the metabolic network by specifying for example that only the elementary modes containing the biomass reaction are needed.

ASP constraints should be expressed this way:

:- not support("biomass"). 

This imposes the solver to have the biomass reaction in the solution.

Alternatively,

:- support("biomass"). 

Imposes the solver to not compute the biomass reaction in the solution.

If you want to use aspefm, we refer you to our paper: Mahout, M., Carlson, R. P. and Peres, S. Answer Set Programming for Computing Constraints-Based Elementary Flux Modes: Application to Escherichia coli Core Metabolism. Processes 8, 1649 (2020).

Additionally, clingo[LP] is property of the University of Potsdam, Germany and collaborators.

Please refer to the following publication: Janhunen, T. et al. "Clingo goes linear constraints over reals and integers." CoRR abs/1707.04053, (2017)).