-*- mode: org -*-/code/moose/branches/asyn13
MOOSE is the Multiscale Object-Oriented Simulation Environment. is the core of a modern software platform for the simulation of neural systems ranging from subcellular components and biochemical reactions to complex models of single neurons, large networks, and systems-level processes.
MOOSE is released under GPLv3.
Homepage: http://moose.ncbs.res.in/ Sourceforge project page: https://sourceforge.net/projects/moose/
==============================================================================
http://sourceforge.net/projects/moose/files/moose/Moose%202.0.0%20Kalakand/
Download and install the prepackaged version suitable for your platform. If no such package exists, you can build it from the sources (download the file ending with tar.gz or tar.bz2 or zip and unpack it, go through README and INSTALLATION).
You can download the development sources using a subversion client: svn co svn://svn.code.sf.net/p/moose/code/moose/branches/async13 moose
or svn checkout svn://svn.code.sf.net/p/moose/code/moose/branches/{branchname} moose
==============================================================================
- g++
- gnu make
- Python development headers and libraries
- GNU Scientific Library (GSL) development packages latest 1.16
- HDF5 development library
- libsbml (if you want SBML support).
- numpy development libraries
- OpenMPI (if you want to run it in parallel).
On Debian based distributions these can be obtained by the command:
sudo apt-get install g++ make python-dev libgsl0-dev python-numpy libhdf5-serial-dev
These names are for Ubuntu 12.04, the exact package names may vary depending on your distribution.
On Fedora based distributions, you can use:
sudo yum install {package names}
where {package names} is the list of Fedora packages corresponding to thos mentioned above.
You may need to inform make of C++ include directories and library directories if your installed packages are at non-standard location. For example, if your have libsbml installed in /opt/libsbml and the header files are located in /opt/libsbml/include and lib files are located in /opt/libsbml/lib, you can set the environment variables CXXFLAGS and LDFLAGS to include these before calling make:
export CXXFLAGS= -I/opt/libsbml/include export LDFLAGS= -L/opt/libsbml/lib
Lookup the manuals for gcc and gnu make to understand CXXFLAGS and LDFLAGS.
cd moose make BUILD=release
cd moose make BUILD=debug
By default, MOOSE is built for Python 2. In case you want to build MOOSE for Python 3K, you need to pass the additional flag:
PYTHON=3
like:
make BUILD=release PYTHON=3
For system-wide installation you can run:
sudo make install
Now you can import moose in a Python script or interpreter with the statement:
import moose
If you have installed the GUI dependencies below for running the graphical user interface, then you can run the GUI by double-clicking on the desktop icon or via the main-menu. The squid axon tutorial/demo is also accessible via these routes.
If you do not have permission to install it in system directories, you can let it be where it was built or copy the `python` subdirectory of MOOSE source tree to a location of your choice and add the path to your PYTHONPATH environment variable. Suppose you have a ~/lib directory where you keep all your locally built libraries, do:
cp -r {moose-source-directory}/python ~/lib/
and add this to your .bashrc file (if you use bash shell):
export PYTHONPATH=”$HOME/lib/python”:”$PYTHONPATH”
For other shells, look up your shell’s manual to find out how to set environment variable in it.
To use the GUI, copy the `gui` subdirectory from MOOSE source tree to a location of your choice, like this:
cp -r {moose-source-directory}}/gui ~/lib/python/moosegui
and make the MooseGUI.py executable:
chmod +x ~/lib/python/moosegui/MooseGUI.py
and create a link somewhere in your PATH:
ln -s ~/lib/python/moosegui/MooseGUI.py ~/bin/moosegui
assuming you have ~/bin directory in your PATH environment variable. After this you run the MOOSE GUI by the `moosegui` command.
In addition to the essential libraries for building MOOSE, you can use many of the Python libraries freely available for scientific computing. Among these are:
- scipy: provides many utility functions for data analysis
- h5py: provides capability to read and write HDF5 files.
To run the moose GUI, you need some additional packages:
- Python bindings for Qt4 or higher
- matplotlib
- Python OpenGL
- Python bindings for Qt’s OpenGL module
On Ubuntu 12.04 these can be installed with:
sudo apt-get install python-matplotlib python-qt4 python-qt4-gl
If you have installed moose as above ‘sudo make install’, and above dependencies, you can run ‘moosegui’ in the terminal, or via the desktop / main menu shortcuts. There are also shortcuts for the squid axon tutorial/demo on the desktop / main menu. MOOSE GUI creates a ~/moose/Demos directory (home folder) on first run. File->load models from there. Help is available from the GUI menu.
==============================================================================
MOOSE can be used as a python module. Look into the Demos directory for sample code. A starting point can be Demos/snippets with useful python code snippets that can be used as building blocks.
MOOSE also comes with a NeuroML reader. Demos/neuroml has some python scripts showing how to load NeuroML models.
MOOSE is backward compatible with GENESIS kinetikit. Demos/Genesis_files has some examples. You can load a kinetikit model with the loadModel function:
moose.loadModel(kkit_file_path, target_model_path)
You can also load GENESIS prototype files. The same loadModel function can be used for this (but you need to have all the channels used in the prototype preloaded in /library):
moose.loadModel(prototype_file_path, prototype_model_path)
Top level moose documentation can be accessed in the Python interpreter the usual way:
import moose
help(moose)
MOOSE classes have built-in documentation that can be accessed via the doc() function:
moose.doc(classname)
will give the full documentation for the class including the fields available.
moose.doc(classname.fieldname)
will give you information about a particular field in a class.
==============================================================================
You can join the MOOSE generic mailing list for your queries: https://lists.sourceforge.net/lists/listinfo/moose-generic
==============================================================================
You can file bug reports and feature requets at the sourceforge tracker: https://sourceforge.net/tracker/?atid=836272&group_id=165660
==============================================================================
Upinder S. Bhalla Primary Architect, Chemical kinetic solvers Niraj Dudani Neuronal solver, parallelization Subhasis Ray Python interface, XML standards G.V.HarshaRani Web page design and User Interface Chaitanya.H Visualization and User Interface Aditya Gilra NeuroML reader and .deb packaging
==============================================================================
MOOSE 2.0.0 “Kalakand”
We announce the release of MOOSE 2.0.0, “Kalakand”*. Debian packages and tarballs can be downloaded from
The latest code is in the usual SourceForge repository at http://sourceforge.net/projects/moose
This is the first full release of MOOSE. In this release the MOOSE team has been conservative about features, and has instead focused on the basics: a decent GUI, documentation, basic numerics, and a cleaner install framework.
In this release you can load and run NEUROML (compartmental neuronal models and networks) from the GUI. You can also load, run, and save chemical kinetic models in the Kinetikit (kkit.g) format. There are demo models for each of these.
There are also a couple of standalone demos, most importantly the Squid demo for the original Hodgkin-Huxley model. This is based on the GENESIS demo of the same name and is an excellent teaching tool.
The scripting language is now Python. The GUI is written in this, and we have production simulations in Python.
Behind the scenes we have set in place a more structured development process with a strong test framework. We invite you to look at the roadmap and indicate what developments you would like best.
We aim to have a much shorter and incremental release cycle from now. Further, the build process is now easier with the debian packaging, so it will be simpler to grab the latest source code from SourceForge if you want to try out new features/bugs hot off the keyboard. Along the way we will fold in the ports to other OSs, depending on what else is prioritized in the roadmap.
*Kalakand is an Indian milk-based sweet. The naming scheme is from an old lab tradition to name things after Indian sweets. The order is alphabetical in the sequence of most Indian alphabets.
The MOOSE team: Niraj Dudani Neuronal solver, parallelization Subhasis Ray Python interface, XML standards G.V.HarshaRani Web page design and User Interface Chaitanya.H Visualization and User Interface Aditya Gilra NeuroML reader and .deb packaging Upinder S. Bhalla Primary Architect, Chemical kinetic solvers