install

聚类分析的源码集

INSTALLATION
============

The installation procedure depends on the software package you want to use.
Below you will find instructions for Cluster 3.0 for Windows, Mac OS X, and
Unix/Linux, Cluster 3.0 as a command-line program, Pycluster (for Python),
and Algorithm::Cluster (for Perl).


Cluster 3.0 for Windows
-----------------------

For Cluster 3.0 for Windows, download the Windows installer from our website
(http://bonsai.ims.u-tokyo.ac.jp/~mdehoon/software/cluster/clustersetup.exe).
Run the installer, and you're done.

If for some reason, you want to recompile Cluster 3.0 from the source, you can
use the Makefile in the windows subdirectory. This makefile was used with
Cygwin/Mingw under Windows, and may need to be modified for other systems.
Type
  make
in the windows subdirectory to compile the C Clustering Library, the Cluster 3.0
GUI, the Windows help files and the documentation. You will need an ANSI C
compiler such as GNU gcc, as well as the GNU windres program to compile the
resources for the GUI. To generate the help files, you will need the HTML Help
SDK, which can be downloaded from Microsoft, as well as the GNU makeinfo
program.

To generate the Windows installer, type
  make clustersetup.exe
For this, you will need the Inno Setup Compiler, which can be downloaded from
http://www.jrsoftware.org.


Cluster 3.0 for Mac OS X
------------------------

Cluster 3.0 can be installed most easily on Mac OS X by using the installer for
Mac OS X, which is available at
http://bonsai.ims.u-tokyo.ac.jp/~mdehoon/software/cluster.

If you want to recompile Cluster 3.0, it is easiest to use Xcode and Interface
Builder that are part of Mac OS X. The subdirectory mac contains the project
file that was used to compile Cluster 3.0.


Cluster 3.0 for Linux/Unix
--------------------------

Cluster 3.0 was ported to Linux/Unix using the Motif libraries. These libraries
are installed on most Linux/Unix computers. You will need a version compliant
with Motif 2.1, such as OpenMotif (http://www.opengroup.org), which is available
at http://www.motifzone.net.

Cluster 3.0 can be installed on Unix/Linux by typing
  ./configure
  make
  make install
This will create the executable cluster and install it in /usr/local/bin. Some
auxiliary files are installed in /usr/local/cluster.


Cluster 3.0 as a command line program
-------------------------------------

To install the command line version of Cluster 3.0, type
  ./configure --without-x
  make
  make install
This will create the executable cluster and install it in /usr/local/bin. Some
auxiliary files (such as the documentation) are installed in /usr/local/cluster.


Python
------

In the top directory (containing setup.py and also this INSTALL file), type
  python setup.py install
You will need a fairly recent version of Python (2.0 or higher) and the
Numerical Python package, as well as an ANSI C compiler such as GNU gcc. You may
need to log in as root the install Pycluster. If you do not have root access, do
  python setup.py install --prefix=/some/other/directory
and set the environment variable
  PYTHONPATH=/some/other/directory/lib/python2.2/site-packages
The exact command depends on which shell you are using; the exact path will
depend on the version of Python you are using.

Note that Pycluster is also available as a tarball containing only the source
code needed for Pycluster. A Windows installer for Pycluster is available from
our website (http://bonsai.ims.u-tokyo.ac.jp/~mdehoon/software/cluster).

Biopython versions 1.11 and up also contain Pycluster. There, it is referred to
as Bio.Cluster. See http://www.biopython.org.


Perl
----

For Perl, type
  perl Makefile.PL
  make
  make test
  make install
You will need Perl version 5.6 or newer. For the install step, you will need
root access. If you do not have root priviliges, use
  perl Makefile.PL prefix=/some/other/directory
to install the module in /some/other/directory/lib/perl5/.
The subdirectory perl/examples contains some example Perl scripts that use
Algorithm::Cluster.



CONTACT
=======

Michiel de Hoon, University of Tokyo, Human Genome Center
[e-mail]