docker.rst

Docker

Build Docker Image

In order to build the image:

sudo docker build --tag <IMAGE_NAME> -f Dockerfile <PATH_TO_DOCKERFILE_DIR>

You can chose any image name (tag) eg. promod.

Run scripts and actions with OST/PM

If script or action requires some external files eg. PDBs, they have to be located in the path accessible via mounted volume and should be accessed via docker (NOT LOCAL) path. Eg. assuming that we have a struc.pdb file in /home/<USER>/pdbs directory and a script.py in /home/<USER> we could mount the /home/<USER> to /home in docker as above by specifying -v /home/<USER>:/home. To run the script we thus need to provide the (relative) path to the script and (relative) path to the file eg:

sudo docker run --rm -v /home/<USER>:/home <IMAGE_NAME> pm script.py pdbs/struct.pdb

or with absolute paths:

sudo docker run --rm -v /home/<USER>:/home <IMAGE_NAME> pm /home/script.py /home/pdbs/struct.pdb

An alternative is to mount the current working directory into the docker home:

sudo docker run --rm -v $(pwd):/home <IMAGE_NAME> pm script.py pdbs/struct.pdb

The Compound Library

At build time of the container, a :class:`~ost.conop.CompoundLib` is generated. Compound libraries contain information on chemical compounds, such as their connectivity, chemical class and one-letter-code. The compound library has several uses, but the most important one is to provide the connectivy information for the rule-based processor.

The compound library is generated with the components.cif dictionary provided by the PDB. As the PDB updates regularly, the compound library shipped with the container is quickly outdated. For most use cases, this is not problematic. However, if you rely on correct connectivity information of the latest and greatest compounds, you have to keep the compound library up to date manually.

The suggested way of doing this is to generate your own compound library and mount it into the container where the original compound lib resides to override it.

The simplest way to create a compound library is to use the :program:`chemdict_tool` available in the container. The program allows you to import the chemical description of the compounds from a MMCIF dictionary, e.g. the components.cif dictionary provided by the PDB. The latest dictionary can be downloaded from the wwPDB site. The files are rather large, it is therefore recommended to download the gzipped version.

After downloading the file use :program:`chemdict_tool` in the container to convert the MMCIF dictionary into our internal format:

sudo docker run --rm -v $(pwd):/home <IMAGE_NAME> chemdict_tool create components.cif.gz compounds.chemlib

To run a script with the upated compound library, use the -v option for mounting/overriding:

sudo docker run --rm -v /home/<USER>:/home -v <COMPLIB_DIR_LOCALHOST>:<COMPLIB_DIR_CONTAINER> <IMAGE_NAME> pm script.py pdbs/struct.pdb

with COMPLIB_DIR_LOCALHOST being the directory that contains the newly generated compound library with name compounds.chemlib and COMPLIB_DIR_CONTAINER the according path in the container. If you didnt change anything in the Dockerfile, the latter should be /usr/local/share/ost_complib

You can check whether the default lib is successfully overriden by looking at the output when running a Python script with following code in the container:

import promod3 # required to setup default lib
from ost import conop
lib = conop.GetDefaultLib()
print lib.GetCreationDate()