Skip to content


Subversion checkout URL

You can clone with
Download ZIP
DNAnexus platform client libraries.
Java C++ Python R Ruby Shell Other
branch: master

DNAnexus Platform SDK

Welcome to the dx-toolkit repository! This repository contains the DNAnexus API language bindings and utilities for interacting with the DNAnexus platform.

See and for relevant documentation.

Installing the toolkit

First, see the section "Build dependencies" below and install the appropriate dependencies for your platform.

To build the toolkit, simply run make.

Then, initialize your environment by sourcing this file:

source dx-toolkit/environment

You will then be able to use dx (the DNAnexus Command Line Client) and other utilities, and you will be able to use DNAnexus API bindings in the supported languages.

Supported languages

The Platform SDK contains API language bindings for the following platforms:

  • Python (requires Python 2.7)
  • C++
  • Perl
  • Java
  • R

Javascript support lives in a separate repo, dnanexus/dx-javascript-toolkit.git.

Build dependencies

The following packages are required to build the toolkit. You can avoid having to install them by either downloading a compiled release from, or by building only a portion of the toolkit that doesn't require them.

Note: There is a known incompatibility (in compiling dxcpp) when using GCC 4.7 with Boost 1.49. Please either use the GCC 4.6 series, or Boost 1.50+.

Ubuntu 14.04

sudo apt-get install make python-setuptools python-pip python-virtualenv python-dev \
  g++ cmake libboost1.55-all-dev libcurl4-openssl-dev zlib1g-dev libbz2-dev flex bison \

Ubuntu 12.04

sudo apt-get install make python-setuptools python-pip python-dev \
  g++ cmake libboost1.48-all-dev libcurl4-openssl-dev zlib1g-dev libbz2-dev flex bison \
sudo pip install --upgrade virtualenv

Ubuntu 10.04

Install Python2.7. Python 2.7 is not available natively on Ubuntu 10.04, but Felix Krull maintains the deadsnakes PPA, which includes a build for Ubuntu 10.04. You can install Python from there as follows (as root):

echo "deb lucid main" > /etc/apt/sources.list.d/deadsnakes.list
apt-key adv --keyserver --recv-keys 5BB92C09DB82666C
apt-get install --yes python2.7 python-pip python-setuptools-deadsnakes
pip-2.7 install virtualenv

Since this repo doesn't contain the Python C headers, you will need to remove the following compiled modules from the dependencies.

sed -i -e '/psutil/ d' src/python/requirements.txt
sed -i -e '/xattr/ d' src/python/requirements_dxfs.txt

The xattr module in particular is needed for the FUSE module. To use the features provided by these modules on Ubuntu 10.04, you will need to arrange for one of the following:

  • Install the Python 2.7 C headers, and then skip the sed steps above to allow the affected modules to be built from source.
  • Install the affected modules (psutil and xattr) in binary form.

Install boost 1.48 or higher (at least the filesystem, program_options, regex, system and thread libraries). This version of boost is not available natively on Ubuntu 10.04. You can use the script build/, which installs it into /usr/local/lib.

The following additional dependencies are also needed:

sudo apt-get install make g++ cmake libcurl4-openssl-dev zlib1g-dev \
  libbz2-dev flex bison autoconf


yum install gcc gcc-c++ automake bison flex python python-pip \
  python-virtualenv boost-devel boost-static cmake openssl-devel \
  libcurl-devel bzip2-devel

This package set was tested on Fedora 20, which has the following package versions (abbreviated list):

  • gcc 4.8.2
  • Python 2.7.5
  • python-pip 1.4.1
  • python-virtualenv 1.10.1
  • boost 1.54.0
  • cmake
  • openssl 1.0.1e
  • libcurl 7.32.0

CentOS/RHEL 5.x/6.x

Install Python 2.7. Python 2.7 is not available natively on CentOS/RHEL 5 or 6. You can use the script build/, which installs it into /usr/local/bin. (Run the script as root.)

Install boost 1.48 or higher (at least the thread and regex libraries). This version of boost is not available natively on CentOS/RHEL 5 or 6. You can use the script build/, which installs it into /usr/local/lib.


yum install cmake libcurl-devel
easy_install-2.7 pip
pip-2.7 install virtualenv


  • On CentOS/RHEL 5.x, two of the utilities, dx-contigset-to-fasta and dx-reads-validator, will not function correctly, as some of the library versions are too old.

  • Tested on CentOS 5.4 and CentOS 6.2.


Install the Command Line Tools for XCode. (Free registration required with Apple)

Install pip and virtualenv for Python:

easy_install-2.7 pip
pip-2.7 install virtualenv

Install the following packages from source or via Homebrew, Fink, or MacPorts:

  • CMake (sudo port install cmake or brew install cmake)
  • Boost >= 1.49 (sudo port install boost or brew install boost)
  • GCC >= 4.6

    • On MacPorts, install and select GCC with:

      sudo port install gcc47
      sudo port select --set gcc mp-gcc47
    • On Homebrew, install and select an up-to-date version of GCC with:

      brew tap homebrew/versions
      brew install gcc47
      export CC=gcc-4.7
      export CXX=g++-4.7
  • bison >= 2.7, autoconf, automake
    • On Homebrew: brew install bison autoconf automake
    • On MacPorts: sudo port install bison autoconf automake


Warning: Not all parts of the SDK are compatible with Windows. Install the following dependencies to build the Upload Agent:

Java bindings

The Java bindings are not built by default. See the Java build documentation for more info.

Javadocs are available on

R bindings

The R bindings (dxR) are not built by default.

Build dependencies

Building and Installing

You should install the RCurl and RJSONIO packages if you have not already. You can do so by running the following commands from within R.

> install.packages("RCurl")
> install.packages("RJSONIO")

Next, run the following command from the command-line (outside of R) to install the package from source.

R CMD INSTALL dx-toolkit/src/R/dxR

You can use the -l flag to specify a local directory in which to install the package if you do not have system permissions to install it for all users. If taking such an approach, you should set your R_LIBS environment variable appropriately if you have not already.

export R_LIBS='/home/username/path/to/dest/library'
R CMD INSTALL dx-toolkit/src/R/dxR -l $R_LIBS

Using the package

Once you have installed the library, you are ready to use it. Inside R, just run:


Ruby bindings

The Ruby package is called dxruby.

Build dependencies

  • Ruby 1.8+
  • rubygems
  • git


Use apt-get install rubygems git to build with Ruby 1.8, or apt-get install ruby1.9.3 git make to build with Ruby 1.9.


On OS X, dependencies may fail to install using Apple Ruby and the XCode toolchain. Instead, use brew install ruby to install the Homebrew Ruby 1.9.


make ruby

Using the package

The environment file will prepend to your GEM_PATH (see Installing the toolkit). To use dxruby, run:

require 'rubygems'
require 'dxruby'

In Ruby 1.9, require 'rubygems' is not necessary.

Upload Agent

See the Upload Agent Readme for Upload Agent build documentation.

Reporting Bugs

Please use GitHub to report bugs, post suggestions, or send us pull requests.

Something went wrong with that request. Please try again.