[bug 2663]

[bioperl-live.git] / INSTALL

# $Id$

Installing Bioperl for Unix

     * 1 BIOPERL INSTALLATION
     * 2 SYSTEM REQUIREMENTS
     * 3 OPTIONAL
     * 4 ADDITIONAL INSTALLATION INFORMATION
     * 5 THE BIOPERL BUNDLE
     * 6 PRELIMINARY PREPARATION
     * 7 INSTALLING BIOPERL THE EASY WAY USING CPAN
     * 8 INSTALLING BIOPERL THE EASY WAY USING 'Build.PL'
     * 9 WHERE ARE THE MAN PAGES?
     * 10 EXTERNAL PROGRAMS

          * 10.1 Environment Variables

     * 11 INSTALLING BIOPERL SCRIPTS
     * 12 INSTALLING BIOPERL IN A PERSONAL MODULE AREA
     * 13 INSTALLING BIOPERL MODULES THE HARD WAY
     * 14 USING MODULES NOT INSTALLED IN THE STANDARD LOCATION
     * 15 THE TEST SYSTEM
     * 16 BUILDING THE OPTIONAL bioperl-ext PACKAGE

          * 16.1 CONFIGURING for BSD and Solaris boxes
          * 16.2 INSTALLATION

BIOPERL INSTALLATION

Bioperl has been installed on many forms of Unix,
Win9X/NT/2000/XP, and on Mac OS X (see the PLATFORMS file for more
details). Following are instructions for installing Bioperl for
Unix/Linux/Mac OS X; Windows installation instructions can be found
in INSTALL.WIN. For installing Bioperl for Mac OS X using Fink, see:

http://www.bioperl.org/wiki/Getting_BioPerl#Mac_OS_X_using_fink

SYSTEM REQUIREMENTS

    * Perl 5.6.1 or later; version 5.8 and greater are highly recommended.

    * External modules: Bioperl uses functionality provided in other
      Perl modules. Some of these are included in the standard perl package
      but some need to be obtained from the CPAN site. The list of external
      modules is included in the DEPENDENCIES file.

OPTIONAL

   * ANSI C or GNU C compiler (gcc) for XS extensions (the
     bioperl-ext package; see BUILDING THE OPTIONAL bioperl-ext
     PACKAGE, below).

ADDITIONAL INSTALLATION INFORMATION

   * Additional information on Bioperl and MAC OS:
      * OS 9 - http://bioperl.org/Core/mac-bioperl.html
      * OS X - http://www.tc.umn.edu/~cann0010/Bioperl_OSX_install.html
            (outdated, but useful for libgd installation notes)
      * OS X - Installing using Fink (in Getting BioPerl)

THE BIOPERL BUNDLE

Users of previous versions of Bioperl may remember Bundle::BioPerl.
You no longer need to install Bundle::BioPerl. Instead, the normal
installation process will ask you if you'd like to install the
optional external module dependencies that Bioperl has.

A full list of BioPerl dependencies can be found inf the DEPENDENCIES
file included with this distribution.

PRELIMINARY PREPARATION

   This is optional, but regardless of your subsequent choice of installation
   method, it will help to carry out the following steps. They will increase
   the likelyhood of installation success (especially of optional
   dependencies).

     * Upgrade CPAN:

 >perl -MCPAN -e shell
 cpan>install Bundle::CPAN
 cpan>q

     * Install/upgrade Module::Build, and make it your preferred installer:

 >cpan
 cpan>install Module::Build
 cpan>o conf prefer_installer MB
 cpan>o conf commit
 cpan>q

     * Install the expat and libgd libraries by whatever method is
       appropriate for your system. If you install libgd in a non-standard
       location, that is fine: when installing the perl module that needs it
       you will be asked where you installed it.

     * If your expat library is installed in a non-standard location, tell
       CPAN about it:

 >cpan
 cpan>o conf makepl_arg "EXPATLIBPATH=/non-standard/lib EXPATINCPATH=/non-standard/include"
 cpan>o conf commit

INSTALLING BIOPERL THE EASY WAY USING CPAN

You can use the CPAN shell to install Bioperl. For example:

 >perl -MCPAN -e shell
 
Or you might have the cpan alias installed:

 >cpan

Then find the name of the Bioperl version you want:

 cpan>d /bioperl/
 CPAN: Storable loaded ok
 Going to read /home/bosborne/.cpan/Metadata
 Database was generated on Mon, 20 Nov 2006 05:24:36 GMT
 Distribution B/BI/BIRNEY/bioperl-1.2.tar.gz
 Distribution B/BI/BIRNEY/bioperl-1.4.tar.gz
 Distribution S/SE/SENDU/bioperl-1.5.2_100.tar.gz

Now install:

 cpan>install S/SE/SENDU/bioperl-1.5.2_100.tar.gz

If you've installed everything perfectly and all the network connections
are working then you may pass all the tests run in the './Build test' phase.
It's also possible that you may fail some tests. Possible explanations:
problems with local Perl installation, network problems, previously
undetected bug in Bioperl, flawed test script, problems with CGI
script used for sequence retrieval at public database, and so on. Remember
that there are over 800 modules in Bioperl and the test suite is running
more than 12000 individual tests, a few failed tests may not affect your
usage of Bioperl.

If you decide that the failed tests will not affect how you intend to use
Bioperl and you'd like to install anyway do:

 cpan>force install S/SE/SENDU/bioperl-1.5.2_100.tar.gz

This is what most experienced Bioperl users would do. However, if you're
concerned about a failed test and need assistance or advice then contact
bioperl-l@bioperl.org. (You could provide us the detailed results of the failed
test(s): see the `THE TEST SYSTEM' below for information on how to generate such
results.)

INSTALLING BIOPERL THE EASY WAY USING Build.PL

The advantage of this approach is it's stepwise, so it's easy to stop and
analyze in case of any problem.

Download, then unpack the tar file. For example:

 >gunzip bioperl-1.5.2_100.tar.gz
 >tar xvf bioperl-1.5.2_100.tar
 >cd bioperl-1.5.2_100

Now issue the build commands:

 >perl Build.PL
 >./Build test

If you've installed everything perfectly and all the network connections
are working then you may pass all the tests run in the './Build test' phase.
It's also possible that you may fail some tests. Possible explanations:
problems with local Perl installation, network problems, previously
undetected bug in Bioperl, flawed test script, problems with CGI script
using for sequence retrieval at public database, and so on. Remember that
there are over 800 modules in Bioperl and the test suite is running more
than 12000 individual tests, a few failed tests may not affect your usage
of Bioperl.

If you decide that the failed tests will not affect how you intend to use
Bioperl and you'd like to install anyway, or if all tests were fine, do:

 >./Build install

This is what most experienced Bioperl users would do. However, if you're
concerned about a failed test and need assistance or advice then contact
bioperl-l@bioperl.org. (You could provide us the detailed results of the failed
test(s): see the `THE TEST SYSTEM' below for information on how to generate such
results.)

To './Build install' you need write permission in the perl5/site_perl/source
area (or similar, depending on your environment). Usually this will require
you becoming root, so you will want to talk to your systems manager if you
don't have the necessary privileges.

It is also straightforward to install the package outside of the this
standard Perl5 location. See INSTALLING BIOPERL IN A PERSONAL MODULE
AREA, below.

WHERE ARE THE MAN PAGES?

When using Makefile.PL (no longer covered in this documentation), we had
to disable the automatic creation of man pages because this step was
triggering a "line too long" error on some OSs due to shell constraints.
If you want man pages installed use the Build.PL installation process
discussed above.

EXTERNAL PROGRAMS

Bioperl can interface with some external programs for executing analyses.
These include clustalw and t_coffee for Multiple Sequence Alignment
(Bio::Tools::Run::Alignment::Clustalw and
Bio::Tools::Run::Alignment::TCoffee) and blastall, blastpgp, and
bl2seq for BLAST analyses (Bio::Tools::Run::StandAloneBlast), and
to all the programs in the EMBOSS suite (Bio::Factory::EMBOSS).

    Environment Variables

Some modules which run external programs need certain environment
variables set. If you do not have a local copy of the specific executable
you do not need to set these variables. Additionally the modules will
attempt to locate the specific applications in your runtime PATH variable.
You may also need to set an environment variable to tell BioPerl about
your network configuration if your site uses a firewall.

Setting environment variables on unix means adding lines like the
following to your shell *rc file.

   For bash or sh:

 export BLASTDIR=/data1/blast

   For csh or tcsh:

 setenv BLASTDIR /data1/blast

Some environment variables include:

+------------------------------------------------------------------------+
| Env. Variable |                      Description                       |
|---------------+--------------------------------------------------------|
|               |Specifies where the NCBI blastall, blastpgp, bl2seq,    |
|BLASTDIR       |etc.. are located. A 'data' directory could also be     |
|               |present in this directory as well, you could put your   |
|               |blastable databases here.                               |
|---------------+--------------------------------------------------------|
|               |If one does not want to locate the data dir within the  |
|BLASTDATADIR or|same dir as where the BLASTDIR variable points, a       |
|BLASTDB        |BLASTDATADIR or BLASTDB variable can be set to point to |
|               |a dir where BLAST database indexes are located.         |
|---------------+--------------------------------------------------------|
|BLASTMAT       |The directory containing the substitution matrices such |
|               |as BLOSUM62.                                            |
|---------------+--------------------------------------------------------|
|CLUSTALDIR     |The directory where the clustalw executable is located. |
|---------------+--------------------------------------------------------|
|TCOFFEEDIR     |The directory where the t_coffee executable is located. |
|---------------+--------------------------------------------------------|
|               |If you access the internet via a proxy server then you  |
|               |can tell the Bioperl modules which require network      |
|               |access about this by using the http_proxy environment   |
|http_proxy     |variable. The value set includes the proxy address and  |
|               |the port, with optional username/password for           |
|               |authentication purposes                                 |
|               |(e.g. http://USERNAME:PASSWORD@proxy.example.com:8080). |
+------------------------------------------------------------------------+

INSTALLING BIOPERL SCRIPTS

Bioperl comes with a set of production-quality scripts that are
kept in the scripts/ directory. You can install these scripts if you'd
like, simply answer the questions during 'perl Build.PL'.
The installation directory can be specified by:

 perl Build.PL
 ./Build install --install_path script=/foo/scripts
 
By default they install to /usr/bin or similar, depending on platform.

INSTALLING BIOPERL IN A PERSONAL MODULE AREA

If you lack permission to install perl modules into the standard
site_perl/ system area you can configure Bioperl to install itself
anywhere you choose. Ideally this would be a personal perl directory or
standard place where you plan to put all your 'local' or personal perl
modules.

   Example:

 >perl Build.PL --install_base /home/users/dag
 >./Build test
 >./Build install

This tells perl to install all the various parts of bioperl in the desired
place, e.g. creating:

   /home/users/dag/lib/perl5/Bio/Perl.pm

Then in your Bioperl script you would write:

 use lib "/home/users/dag/lib/perl5/";
 use Bio::Perl;

For more information on these sorts of custom installs see the documentation
for Module::Build.

If you are used to using something like:

 >perl Makefile.PL PREFIX=/home/users/dag

You can get similar behaviour by using this instead:

 >perl Build.PL --prefix /home/users/dag

For more information, see Module::Build::Cookbook documentation for
Installing_in_the_same_location_as_ExtUtils::MakeMaker

You can also use CPAN to install modules in your local directory. First
enter the CPAN shell, then set the arguments for the commands
"perl Makefile.PL" and "./Build install", like this:

 >perl -e shell -MCPAN
 cpan>o conf makepl_arg LIB=/home/users/dag/My_Local_Perl_Modules
 cpan>o conf mbuild_install_arg "--install_path lib=/home/users/dag/My_Local_Perl_Modules"
 cpan>o conf commit

INSTALLING BIOPERL MODULES THE HARD WAY

As a last resort, you can simply copy all files in Bio/ to any directory
in which you have write privileges. This is generally NOT recommended
since some modules may require special configuration (currently none do,
but don't rely on this).

You will need to set "use lib '/path/to/my/bioperl/modules';" in your perl
scripts so that you can access these modules if they are not installed in
the standard site_perl/ location. See above for an example.

To get manpage documentation to work correctly you will have to
configure man so that it looks in the proper directory. On most systems
this will just involve adding an additional directory to your $MANPATH
environment variable.

The installation of the Compile directory can be similarly redirected, but
execute the make commands from the Compile/SW directory.

If all else fails and you are unable to access the perl distribution
directories, ask your system administrator to place the files there for
you. You can always execute perl scripts in the same directory as the
location of the modules (Bio/ in the distribution) since perl always
checks the current working directory when looking for modules.

USING MODULES NOT INSTALLED IN THE STANDARD LOCATION

You can explicitly tell perl where to look for modules by using the
Lib module which comes standard with perl.

   Example:

 #!/usr/bin/perl
 use lib "/home/users/dag/lib/perl5/";
 use Bio::Perl;
 #<...insert whizzy perl code here...>

Or, you can set the environmental variable PERL5LIB:

   csh or tcsh:

 setenv PERL5LIB /home/users/dag/lib/perl5/

   bash or sh:

 export PERL5LIB=/home/users/dag/lib/perl5/

THE TEST SYSTEM

The Bioperl test system is located in the t/ directory and is
automatically run whenever you execute the './Build test' command
(having previously run 'Perl Build.PL'; if you have already installed
Bioperl answer 'no' to script installation to get nicer test output
later).
Alternatively if you want to investigate the behavior of a specific test
such as the Seq test you would type:

 >./Build test --test_files t/Seq.t --verbose

The ./ ensures you are using the Build script in the current directory to
make sure you are testing the modules in this directory not ones
installed elsewhere. The --test_files arguement can be used multiple times
to try a set of test scripts in one go. The --verbose arguement outputs
the detailed test results, instead of just the summary you see during
'./Build test'.

If you are trying to learn how to use a module, often the test suite is a
good place to look. All good extreme programmers try and write a test
BEFORE they write the module to insure that their module behaves the way
they expect. You'll notice some 'ok' and 'skip' commands in a test, this
is part of the Perl test suite that signifies a passed test with an 'ok
N', where N is the test number. Alternatively you can tell Perl to skip
tests. This is useful when, for example, your test detects that the
network is not present and thus should skip, not fail, any tests that
require a network connection.

BUILDING THE OPTIONAL bioperl-ext PACKAGE

The bioperl-ext package contains C code and XS extensions for
various alignment and trace file modules (Bio::Tools::pSW for DNA
Smith-Waterman, Bio::Tools::dpAlign for protein Smith-Waterman,
Bio::SearchDist for EVD fitting of extreme value,
Bio::SeqIO::staden).

This Installation may work out-of-the box for most platforms except BSD
and Solaris boxes. For other platforms skip this next paragraph.  Of note,
the code for bioperl-ext has not been updated along with the rest of bioperl,
so one may expect to see some issues.  If so, please report them to the
BioPerl mailing list.

    CONFIGURING for BSD and Solaris boxes

You should add the line -fPIC to the CFLAGS line in
Compile/SW/libs/makefile. This makes the compile generate position
independent code, which is required for these architectures. In addition,
on some Solaris boxes, the generated Makefile does not make the correct
-fPIC/-fpic flags for the C compiler that is used. This requires manual
editing of the generated Makefile to switch case. Try it out once, and if
you get errors, try editing the -fpic line

    INSTALLATION

Move to the directory bioperl-ext. This is available as a separate package
released from ftp://bioperl.org/pub/bioperl/DIST. This is where the C
code and XS extension for the bp_sw module is held and execute these
commands: (possibly after making the change for BSD and Solaris, as
detailed above)

 perl Makefile.PL   # makes the system specific makefile
 make          # builds all the libaries
 make test     # runs a short test
 make install  # installs the package correctly.

This should install the compiled extension. The Bio::Tools::pSW
module will work cleanly now.