Mitochondria are organelles in biological cells essential for keeping the cells alive. They are able to move around the cytosol (intracellular volume outside nucleus), where they form extensive rapidly reorganizing networks. Quantitative description of their morphology and dynamics was problematic because of the seemingly irregular and constantly changing reticular conformations they can adopt. This problem is tackled here by representing the mitochondria as a graph evolving in time.
The graph evolution results from the division and fusion of branches, implemented as a
time-dependent stochastic process. Specialized protein complexes performing these
transformations in the mitochondria are known from empirical studies, but the dynamics is not
dependent on the details of their operation and may be implemented in more general terms.
In this minimal representation, node degrees are constrained to a very narrow range (1, 2 and 3),
and only the graph topology is accounted for, thus neglecting the embodiment of the network in space.
Such formulation corresponds to a well-mixed chemical system, so that any graph node has
equal probability to interact with any other node appropriate for the reaction type.
The evolution corresponds to a minimization of the free energy of node interactions.
So, in the coarse of the simulation run, the network settles in a dynamic steady-state configuration,
defined by the reaction parameters and the network dimensions.
Formally, the process corresponds to a solution of the master equation describing the
transformations in continuous time.
More details can be found in the original
publication:
Sukhorukov VM, Dikov D, Reichert AS, Meyer-Hermann M. Emergence of the mitochondrial
reticulum from fission and fusion dynamics.
PLoS Comput Biol. 2012, 8: e1002745-10.1371/journal.pcbi.1002745.
Please reference the above manuscript whenever the results produced by this code are made publicly available.
The simulator makes use of the general-purpose library utils, (included here as a git submodule). So, when cloning, please make sure, that the dependent library is included, e.g., by applying:
git clone --recursive https://github.com/vsukhor/mitoSim.gitand, in the case of an update, do this explicitly on the submodule:
git submodule init
git submodule updatemitoSim requires a C++20 - capable compiler. It was tested on macOS (clang-12) and Ubuntu (gcc-11). For generation of pseudo-random numbers, utils::random relies on either boost or NVIDIA cuRAND, so depending on the underlying generator chosen, one of these should be accessible (in the case of the former, the headers suffice).
Using cmake (ver. 3.15 or higher):
cd mitosim
cmake -S . -B build
cmake --build build The simulation parameters are read from a short configuration file, structured as in the example config_sample.txt. More details on the config file formatting can be found in the Utils::Config documentation The file is expected to be named as 'config_X.txt', where X is the configuration-specific suffix and the rest of the name is fixed. Then, using the above example, and the executable 'mitosim' the simulation may be launched as
./mitosim /path/to/configfile/ sample 22 28The first two arguments are the working directory where the config file is positioned and the config file suffix respectively. The last two arguments are the range limits of the run indexes used for choosing rng seeds.
Python 3 code for basic visualization and analysis of simulation results can be found in mitosim-postProcessor. The scripts can be used independently of the main C++ code, provided the output files are available (see the instructions).