Wormtable is a write-once read-many table for large scale datasets. It provides Python programmers with a simple and efficient method of storing, processing and searching datasets of essentially unlimited size. A wormtable consists of a set of rows, each of which contains values belonging to a fixed number of columns. Rows are encoded in a custom binary format, designed to be flexible, compact and portable. Rows are stored in a data file, and the offsets and lengths of these rows are stored in a Berkeley DB database to support efficient random access. Wormtable also supports efficient searching and retrieval of rows with particular values through the use of indexes, also based on Berkeley DB.
The Variant Call Format (VCF) is supported directly by wormtable through a command line conversion program, vcf2wt. There is also a command line utility wtadmin to manage wormtables, including the ability to dump values and add, remove and view indexes.
If you use wormtable in your work, please cite the BMC Bioinformatics
CITATION.txt file for details.
Full documentation for
wormtable is available at
Quick install for Debian/Ubuntu
If you are running Debian or Ubuntu, this should get you up and running quickly:
$ sudo apt-get install python-dev libdb-dev $ sudo pip install wormtable
For Python 3, use
Once Berkeley DB has been installed (see below) we can build the
wormtable module using the
standard Python methods. For
example, using pip we have
$ sudo pip install wormtable
Or, we can manually download the package, unpack it and then run:
$ python setup.py build $ sudo python setup.py install
Most of the time this will compile and install the module without difficulty.
It is also possible to download the latest development version of
wormtable from github.
Wormtable requires the
argparse package, which was introduced to the
standard library for version 3.2 (it is also included in 2.7). For users
of older Python versions, the
argparse module must be installed for
the command line utilities to work:
$ sudo pip install argparse
This is not necessary for recent versions of Python.
Installing Berkeley DB
Wormtable requires Berkeley DB (version 4.8 or later), which is available for all major platforms.
Installing Berkeley DB is very easy on Linux distributions.
On Debian/Ubuntu use:
$ sudo apt-get install libdb-dev
and on Red Hat/Fedora use:
# yum install libdb-devel
Other distributions and package managers should provide a similarly easy option to install the DB development files.
Mac OS X
For MacPorts, to install e.g. v5.3
$ sudo port install db53
Then, to build/install wormtable, we need to set the CFLAGS and LDFLAGS environment variables to use the headers and libraries in /opt:
$ CFLAGS=-I/opt/local/include/db53 LDFLAGS=-L/opt/local/lib/db53/ python setup.py build $ sudo python setup.py install
For Homebrew, get the current Berkeley DB version and again build wormtable after setting CFLAGS and LDFLAGS appropriately:
$ brew install berkeley-db $ CFLAGS=-I/usr/local/Cellar/berkeley-db/5.3.21/include/ LDFLAGS=-I/usr/local/Cellar/berkeley-db/5.3.21/lib/ python setup.py build $ sudo python setup.py install
For more details of Berkely DB versions, see here: https://www.macports.org/ports.php?by=category&substr=databases
On platforms that Berkeley DB is not available as part of the native packaging
system (or DB was installed locally because of non-root access)
there can be issues with finding the correct headers and libraries
wormtable. For example,
if we add the DB 4.8 package on FreeBSD using:
# pkg_add -r db48
we get the following errors when we try to install wormtable:
$ python setup.py build ... [Messages cut for brevity] ... _wormtablemodule.c:3727: error: 'DB_NEXT_NODUP' undeclared (first use in this function) _wormtablemodule.c:3733: error: 'DB_NOTFOUND' undeclared (first use in this function) _wormtablemodule.c:3739: error: 'DistinctValueIterator' has no member named 'cursor' _wormtablemodule.c:3739: error: 'DistinctValueIterator' has no member named 'cursor' _wormtablemodule.c:3740: error: 'DistinctValueIterator' has no member named 'cursor' error: command 'cc' failed with exit status 1
This is because the compiler does not know where to find the headers and library
files for Berkeley DB.
To remedy this we must set the
CFLAGS environment variables to
their correct values. Unfortunately there is no simple method to do this
and some knowledge of where your system keeps headers and libraries
is needed. To complete the installation for the FreeBSD example above,
we can do the following:
$ CFLAGS=-I/usr/local/include/db48 LDFLAGS=-L/usr/local/lib/db48 python setup.py build $ sudo python setup.py install
Installation without root access
If you need to install wormtable on a system where Berkeley DB is not installed (and your system administrator refuses to install it, for some reason), we can still compile and install it locally. Here is a recipe that worked on a Debian squeeze machine; however, this is not guaranteed to work on any given system and you may need to tweak things a little to suit your environment:
$ mkdir -p $HOME/.local $ wget http://download.oracle.com/berkeley-db/db-4.8.30.tar.gz $ tar -zxf db-4.8.30.tar.gz $ cd db-4.8.30/build_unix/ $ ../dist/configure --prefix=$HOME/.local $ make install
This downloads a version of Berkeley DB from Oracle, compiles and then installs it to the directory $HOME/.local. (The version of Berkeley DB you use doesn't really matter once it's at least 4.8.) Now, download the latest version of wormtable, untar it and cd to the new directory. We can then install it locally:
$ CFLAGS=-I$HOME/.local/include LDFLAGS=-L$HOME/.local/lib/ python setup.py install --user
Now we need to set up some paths so that we can use this at run time. Put the following lines into your $HOME/.bashrc (or equivalent if you use another shell):
export LD_LIBRARY_PATH=$HOME/.local/lib:$LD_LIBRARY_PATH export PATH=$HOME/.local/bin:$PATH
Then, log out, log back in, and you should be able to use wormtable.
Wormtable has an extensive suite of tests to ensure that data is stored correctly. It is a good idea to run these immediately after installation:
$ python tests.py
Wormtable is highly portable, and has been successfully built and tested on the following platforms:
|Ubuntu 13.04||x86-64||2.7.4||gcc 4.7.3|
|Ubuntu 13.04||x86-64||3.3.1||gcc 4.7.3|
|Ubuntu 13.04||x86-64||2.7.4||clang 3.2.1|
|Debian squeeze||x86-64||2.6.6||gcc 4.4.5|
|Debian squeeze||x86-64||3.1.3||gcc 4.4.5|
|Debian squeeze||x86-64||3.1.3||clang 1.1|
|Debian squeeze||ppc64||2.6.6||gcc 4.4.5|
|Debian squeeze||ppc64||3.1.3||gcc 4.4.5|
|Debian wheezy||armv6l||2.7.3||gcc 4.6.3|
|Fedora 17||i386||2.7.3||gcc 4.7.2|
|Fedora 17||i386||3.2.3||gcc 4.7.2|
|FreeBSD 9.0||i386||3.2.2||gcc 4.2.2|
|FreeBSD 9.0||i386||2.7.2||gcc 4.2.2|
|FreeBSD 9.0||i386||3.1.4||clang 3.0|
|OS X 10.8.4||x86-64||2.7.2||clang 4.2|
|Solaris 10||SPARC||3.3.2||gcc 4.8.0|
|Solaris 11.1||SPARC||2.6.8||gcc 4.5.2|
|Solaris 11.1||SPARC||2.6.8||Sun C 5.12|
|Scientific Linux 6.2||x86-64||2.6.6||icc 12.0.0|