The Readme

Download:From sourceforge or follow the Quick installation instructions.
Registration:To be informed of bugs, new releases please subscribe to the mailing lists at sourceforge.

Dependencies

The toolkit requires Python 2.4.3 or greater, and Numpy 1.0 or greater. Aside from these the dependencies below are optional and the code will work as is. A C compiler, however, will allow external C module’s responsible for the likelihood and matrix exponentiation calculations to be compiled, resulting in significantly improved performance.

Note

While Numpy 1.0 is supported, we encourage users to use the latest stable Numpy release (currently 1.3) since we will be upgrading the required version soon.

Required

  • Python: the language the toolkit is primarily written in, and in which the user writes control scripts.

  • Numpy: This is a python module used for speeding up matrix computations. It is available as source code for *nix.

    Note

    For installing Numpy, unless you know what you are doing we recommend not linking Numpy against ATLAS or LAPACK libraries. On systems where these have been installed, this can be achieved by setting shell environment variables prior to building Numpy as

    $ export ATLAS=None
    $ export LAPACK=None

Optional

  • C compiler: This is standard on most *nix platforms. On Macos X this is available for free in the Developer tools which, if you don’t already have them, can be obtained from Apple.
  • ReportLab: required for drawing trees and alignments to pdf.
  • Matplotlib: used to plot several kinds of graphs related to codon usage.
  • Cython: This module is only necessary if you are a developer who wants to modify the *.pyx files.
  • PyxMPI: Our own python MPI interface, required for parallel computation.
  • SQLAlchemy and MySQL-python : These are required for the Ensembl querying code.

If you use the Quick installation approach, these are all specified in the requirements file.

Installation

If you don’t wish to use the Quick installation approach then the conventional *nix platform (including MacOS X) python package installation procedure applies. Download the software from here. Uncompress the archive and change into the PyCogent directory and type:

$ python setup.py build

This automatically compiles the modules. If you have administrative privileges type:

$ sudo python setup.py install

This then places the entire package into your python/site-packages folder.

If you do not have administrator privileges on your machine you can change the build approach to:

$ python setup.py build_ext -if

which compiles the extensions in place (the i option) forcibly (the f option, ie even if they’ve already been compiled). Then move the cogent directory to where you want it (or leave it in place) and add this location to your python path using sys.path.insert(0, "/your/path/to/PyCogent") in each script, or by setting shell environment variables (e.g. $ export PYTHONPATH=/your/path/to/PyCogent:$PYTHONPATH)

Testing

PyCogent/tests contains all the tests (currently >3100). You can most readily run the tests using the PyCogent/run_tests shell script. This is done by typing:

$ sh run_tests

which will automatically build extensions in place, set up the PYTHONPATH and run PyCogent/tests/alltests.py. Note that if certain optional applications are not installed this will be indicated in the output as “can’t find” or “not installed”. A “.” will be printed to screen for each test and if they all pass, you’ll see output like:

Ran 3299 tests in 58.128s

OK

Tips for usage

A good IDE can greatly simplify writing control scripts. Features such as code completion and definition look-up are extremely useful. For a complete list of editors go here.

To get help on attributes of an object in python, use

>>> dir(myalign)

to list the attributes of myalign or

>>> help(myalign.writeToFile)

to figure out how to use the myalign.writeToFile method. Also note that the directory structure of the package is similar to the import statements required to use a module – to see the contents of alignment.py or sequence.py you need to look in the directory cogent/core path, to use the classes in those files you specify cogent.core for importing.

Table Of Contents

Previous topic

Quick installation

Next topic

Coding guidelines

This Page