NAME¶
getData - retrieves databases from the Internet
SYNOPSIS¶
getData [ --mirrordir <path> ] <list of db names>
getData --list
DESCRIPTION¶
Bioinformatics has the intrinsic problem to bring the biological data to the end
user. Astronomers have the equivalent problem and particle physicists, well,
they haven come up with (first) the web and (second) the computational grids
to address their problems. Debian helps with the programs but will not provide
such huge datasets that are even frequently updated - not even in
volatile.debian.org. Most bioinformatics researchers will not need too many of
such databases. And even more so will gladly continue in using public services
remotely.
For those who need a set of databases on a regular basis, this script shall be a
start to automate the burden to download the data and update indices and the
like. The world has seen such magic before with the Lion Biosciences Prisma
tool (
http://bib.oxfordjournals.org/cgi/reprint/3/4/389.pdf) but how about
something simpler (as a start) that at least gets close to what we desire and
is Free. The aim must be to address the needs of all (most) communities, not
only of the bioinformatics world. The seed was hence made with databases from
astronomy.
Please contact the Debian-Med community if you consider this program to be
almost ready for your needs and explain what still needs to be added. Public
databases that you managed to integrate with this system are also very warmly
welcomed as feedback.
OPTIONS¶
- --help
-
this help
- --man
- Present a more detailed description in form of a man
page.
- --verbose
- Say one or two words more than required.
- --mirrordir <path>
- Specifies destination directory. The data will be mirrored
to the folder $mirrordir/$dbname/. Please be aware that this mirrordir is
nowhere stored. The directory can consequently be moved to arbitrary
locations at any time, if the users of the data are only informed about
that moving.
- --list
- Lists all databases that may be requested to be
installed.
- <list of db names>
- Only those databases that are explicitly requested to be
downloaded will be downloaded. Such databases may require considerable
bandwidth, so please make sure you know you are doing the right
thing.
- --post
- Perform only the unpacking/indexing, but do not
retrieve/update the databases. This option is considered useful when
adding a new database management system to the system, e.g. after
installing EMBOSS.
- --source
- Perform only the unpacking/indexing, but do not
retrieve/update the databases. This option may be beneficial when the site
administator is aware of current analyses that should not be disturbed by
the indexing process but the downloading from the net can already be
started.
- --confd <directory>
- Allows for the specification of a directory in which
multiple files can be stored that will be read by getData upon its
invocation. These may add values to the global variable %toBeMirrored that
specifies the databases and their download scripts.
- --config <system>
- Preparation of the configuration file that would be reuired
for a particular system that deals with the database. The configuration is
printed to stdout and is expected to be copied manually to the proper file
or folder. One could imagine this process to be automated, though this is
not yet implemented. Currently available is support for two systems:
- emboss
- This specifies the EMBOSS suite of tools for bioinformatics
(www.emboss.org) that is also available as a Debian package. The
configuration for the Uniprot databases will allow the sequence retrieval
with the seqret tool.
- dre - ARC Grid Runtime Environment
- Runtime environments (REs) are a concept of the ARC grid
middleware of which more can be learned on http://www.nordugrid.org. A
script is needed to indicate the presence of a runtime environment. Here,
the name of the script is important, which is not definable by getData
though since it only writes to stdout.
Unfortunately, the configuration was not yet be found to be modularised. It all
needs to happen within the getData script itself.
- --remove <list of dbnames>
- This command removes folders that store the data. In
principle this could be perfomed manually, though some databases may have
special requirements pre- or post-removal, which can be specified
individually for every database.
SPECIFICATION OF DATABASES¶
Databases for download and their post-processing are specified at two different
locations. One is the getData script itself, the other are files stored in
/etc/getData.d. Either will define elements of a considerably large hash. The
key is the identifier which is also shown by the 'getData --list' directive.
The value is a reference to another hash, which assigns values to all the
properties that a database has for its download and post-processing:
- name - a human-readable pretty-printed name or short
description that makes clear to the world what this database is about.
- A bad example is the mere assignment of "DE405",
which few people understand. A better example is "Pfam-A : Manually
curated protein families and domains, only the seed is presented.".
One could argue that one should have that field renamed to
"description".
- source - shell commands to perform the initial download and
subsequent updates
- Commonly the wget tool is used for download. The such
presented little script is executed underneath the mirrordir directory.
One simple example is "wget --mirror
ftp://ssd.jpl.nasa.gov/pub/eph/export/unix/unxp2[01]*.405". With
increasing proficiency in using wget, one is tempted to substitute
"--mirror" with "--recursive --no-host-directories
--no-directories --level 1 --no-parent".
- post-download - shell commands to perform after the data
has been downloaded.
- A simple (and unnecessary when used the right flags to
wget) example is the mere setting of a symbolic link:
"post-download" => "ln -s ssd.jpl.nasa.gov/pub/eph/export/unix/unxp*.405 ."
Some more effort has been put into TrEMBL for the merging of releases with
subsequent updates and the indexing for EMBOSS:
"d=uncompressed; if [ ! -d \$d ]; then mkdir \$d; fi; "
."rm -rf \$d/trembl.dat; "
."(find ftp.ebi.ac.uk -name '*.dat.gz' | xargs -r zcat ) > \$d/trembl.dat; "
."[ -x /usr/bin/dbxflat ] "
. "&& cd \$d && "
. "dbxflat -dbresource embl -dbname trembllocal -idformat swiss -filenames=trembl.dat -fields id,acc -auto",
The dots are connecting strings in Perl. This helps the readability of the
code. When writing these scripts, please be aware the newlines don't
separate the individual commands here. Semicolon are required.
- recommends - suggests a series of packages to be present
for the use of the database or the performance of the indexing.
- This information is not used at the moment, also to render
this script more useful for other Linux distributions than Debian.
- getWgetOptions - private command to get wget options
- This is used at download time by makefiles, is not intended
to be used interactively, and could be removed anytime.
EXAMPLES¶
The following will list the identifiers and the descriptions of the first 4
databases that area available via getData on your system.
./getData --mirrordir=/local/databases/mirrored --list | head 4
To install any particular database, only give its name as an argument. If the
installation is performed at another directory than the default, then the
--mirrordir needs again to be set.
./getData swiss.dat
To remove the database again, give the script a hint with the --remove flag
./getData --remove swiss.dat
To perform the indexing only and circumvent the download (attention, this is
dangerous since the index files will look newer than the database is), do
./getData --post swiss.dat
A special exception to these extra scripts is the --config flag in that it takes
a list of extra arguments. Each shall denote a particular system that this
database may be of interest for. There are today two systems supported:
TODO¶
We now need a mechanism with which packages can specify hooks that shall be
called upon an update of a database. But we cannot assume that every indexing
that can be performed because of the installation of some package is also
desired by the user. How to configure this properly is left to be decided.
SEE ALSO¶
http://debian-med.alioth.debian.org, http://wiki.debian.org/DebianMed,
/etc/getData.conf
AUTHORS¶
This script was prepared by Steffen Moeller <moeller@debian.org> and
Charles Plessy <debian-no-spam@plessy.org> and is distributed under the
terms of the GNU Public License (GPL). On Debian systems, this license can be
found under /usr/share/common-licenses/GPL.