The available tags are:
- latest (default): NCBI BLAST+ 2.2.30+ (aliased to 2.2.30plus)
- 2.2.30plus: NCBI BLAST+ 2.2.30+
- 2.2.29plus: NCBI BLAST+ 2.2.29+
- 2.2.28plus: NCBI BLAST+ 2.2.28+
- 2.2.27plus: NCBI BLAST+ 2.2.27+
- 2.2.26plus: NCBI BLAST+ 2.2.26+
Installation and Use
Installation and use of a tool wrapper (links to follow) specifying one of these images will automatically pull and run (assuming that Galaxy's job_conf.xml has been set up to use Docker, and Docker is installed)
docker pull simonalpha/ncbi-blast-docker
Specific versions can be specified via tagging, for instance:
docker pull simonalpha/ncbi-blast-docker:2.2.30plus
Run an instance of the image, mounting in directories with the required sequences/databases/files and setting the working directory as appropriate. If no program is specified, the default behaviour is to drop you into a shell in your working directory.
docker run --rm -v /home/simonalpha/project:/blast -v /databases:/db -w /blast simonalpha/ncbi-blast-docker
All the BLAST binaries in the default distribution are available and on the path, and you can run BLAST as usual.
Another option is to provide a BLAST invocation when starting a container; much more useful for scripting! Ensure your BLAST invocation uses the paths you are mounting to in the container.
docker run --rm -v /home/simonalpha/project:/blast -v /databases:/db -w /blast simonalpha/ncbi-blast-docker blastp -query /blast/q_seq.faa -db /db/prot_db -out q_seq_V_prot_db
The above is based strongly on how Galaxy uses Docker containers, however other methods are definitely possible. If you come across any, please do let us know.
I recommend use of
--rm in your run commands, it automatically clears the containers after use. However, this requires you write your data into a mounted path, otherwise your results will be deleted with the container.
Users and Permissions
These images currently run as root, to allow easy read-write access to mounted volumes. This will change in the future with enabling patches in Galaxy, based around UIDs rather than users.
The layout of this repository is based on the structure of the Java Dockerfile repository for clear separation of versions and easy setup of automated builds. Suggestions or Pull Requests for better organisation most welcome.