Conda Build Recipes

DAlphaBall is distributed under the GNU LGPL. A build recipe and associated scripting components for building conda packages for Rosetta components. These recipes can be used to generate the DAlphaBall executable.

TLDR

1. Install docker.
2. Use dalphaball_docker_build.sh <output dir> to build the dalphaball conda package.

Building on/for macOS (without Docker)

The c std lib that can be used with conda build is a little older, so you'll need to set up an appropriate build environment. The Mac OS X 10.10 SDK is compatible and can be configured to work with conda build with the following steps:

1. Get MacOSX10.10.sdk
2. Move the SDK to /opt with tar xf ~/Downloads/MacOSX10.10.sdk.tar.xz -C /opt
3. Add the following to ~/.condarc:

conda_build:
  config_file: ~/.conda/conda_build_config.yaml

4. Create ~/.conda/conda_build_config.yaml if it doesn't exist and add:

CONDA_BUILD_SYSROOT:
  - /opt/MacOSX10.10.sdk # [osx]

5. Create and activate a new conda environment (or activate an existing conda environment), installing conda-build and pyrosetta packages. For older versions of conda (<=4.6), one may need to install conda-build<=3.22.0 to avoid an ImportError at the next stage.
6. Navigate to this cloned repository directory, and use conda build recipes -c conda-forge --croot=<output dir> to build the dalphaball conda package.
7. In order to use the dalphaball package with pyrosetta, then use conda install dalphaball -c <output dir> to install dalphaball into the active conda environment. Optionally, it may be installed into any conda environment at this stage.
8. The dalphaball executable is then located in the active conda environment bin directory, usually located here: ~/opt/anaconda3/envs/MYENV/bin/dalphaball.
9. If using Jupyter notebooks, launch jupyter notebook from the active conda environment. The conda environment bin directory should be in $PATH, so one may initialize PyRosetta with the following flag: -holes:dalphaball dalphaball. If jupyter notebook is launched from a different conda environment, make sure ipykernel is installed into the active conda environment and the active conda environment is registered for use as a kernel on the Jupyter server, then specify the full path to the dalphaball executable during PyRosetta initialization: -holes:dalphaball ~/opt/anaconda3/envs/MYENV/bin/dalphaball.
10. Enable usage of dalphaball in the BuriedUnsatHbonds filter with the dalphaball_sasa="1" option. The Holes filter automatically uses it.

Build Layout

These components are roughly organized into layers:

• recipes: Conda recipe definitions with dependency information and build steps.

• dalphaball_docker_build.sh: Script using minimal build environment defined in linux-anvil to build broadly compatible linux conda packages. This should be considered the primary entrypoint to generate conda packages.

• Use conda build recipes -c conda-forge --croot=<output dir> to build the dalphaball conda package in your current environment.

Build Debugging

The dalphaball_docker_build.sh build scripts can be used to diagnose failed builds within the anvil environment. To debug a failed build:

• Note the build prefix in the container. Eg.

  /home/conda/build/dalphaball_12345

• Re-invoke the docker run config used to execute the build, replacing the conda build call with a direct invocation of the workspace build script.

  Eg. bash -c 'cd /home/conda/root/dalphaball && conda build recipes -c conda-forge --croot /home/conda/build' -> bash -c 'cd /home/conda/build/dalphaball_12345 && ./conda_build.sh'