A spatially-explicit neutral ecology simulator using coalescence methods
A python package for coalescence-based spatially-explicit neutral ecology simulations
pycoalescence is a python package for spatially-explicit coalescence neutral simulations. pycoalescence provides a pythonic interface for setting up, running and analysing spatially-explicit neutral simulations. Simulations themselves are performed in c++ using necsim for excellent performance. Usage of python allows for easier setup of simulation parameters.
For full documentation please see here.
Installation is available through pip, conda or a manual installation. For full installation instructions, see here.
Currently, pip is supported on Mac OS X and Linux and conda is supported on Linux and Windows.
Using pip, make sure all the prerequisites are installed and run pip install pycoalescence.
If you cannot install via pip, download the tar ball and run python setup.py install. The package can also be installed locally, (i.e not to the virtual or system environment) using python installer.py in the module directory. Either method requires all dependencies have been installed. By default, .o files are compiled to lib/obj and the .so file is compiled to the necsim directory.
Make sure compilation is performed under the same python version simulations will be performed in.
The Simulation class contains most of the operations required for setting up a coalescence simulation. The important set up functions are:
- set_simulation_params() sets a variety of key simulation variables, including the seed, output directory, dispersal parameters and speciation rate.
- set_map() is used to specify a map file to use. More complex map file set-ups can be provided using set_map_files. set_map_parameters() can also be used to customise parameters, instead of detecting from the provided tif files.
- set_speciation_rates() takes a list of speciation rates to apply at the end of the simulation. This is optional.
- run() checks and starts the simulation, writing to the output database upon successful completion. This stage can take an extremely long time (up to tens of hours) depending on the size of the simulation and the dispersal variables. Upon completion, an SQL file will have been created containing the coalescence tree.
The CoalescenceTree class also contains some basic analysis abilities, such as applying additional speciation rates post-simulation, or calculating species abundances for fragments within the main simulation.
The basic procedure for this procedure is
- set_database() to provide the path to the completed simulation database
- set_speciation_params() which takes as arguments
- T/F of recording full spatial data
- either a csv file containing fragment data, or T/F for whether fragments should be calculated from squares of continuous habitat.
- list of speciation rates to apply
- [optional] a sample file to specify certain cells to sample from
- [optional] a config file containing the temporal sampling points desired.
- apply() performs the analysis. This can be extremely RAM and time-intensive for large simulations. The calculations will be stored in extra tables within the same SQL file as originally specified.
- Python version 2 >= 2.7.9 or 3 >= 3.4.1
- C++ compiler (such as GNU g++) with C++14 support.
- The SQLite library available here. Requires both c++ and python installations. Comes as standard with python.
- The Boost library for C++ available here.
- Numerical python (numpy) package (pip install numpy).
- The gdal library for both python and C++ (available here). Although it is possible to turn off gdal support, this is not recommended as it is essential if you wish to use .tif files for simulation. Both the python package and c++ binaries are required; installation differs between systems, so view the gdal documentation for more help installing gdal properly.
- The fast-cpp-csv-parser by Ben Strasser, available here. This provides much faster csv read and write capabilities and is probably essential for larger-scale simulations, but not necessary if your simulations are small. The folder fast-cpp-csv-parser/ should be in the same directory as your necsim C++ header files (the lib/necsim directory).
- Scipy package for generating fragmented landscapes (pip install scipy).
- Matplotlib package for plotting fragmented landscapes (pip install matplotlib).