-
Notifications
You must be signed in to change notification settings - Fork 2
KaBOB installation (AllegroGraph)
To build KaBOB using the AllegroGraph triple-store you will need to obtain an AllegroGraph license from Franz Inc. Note: The triple limit for the free evaluation license provided by Franz is too restrictive to build a KaBOB instance. A full license is required to use AllegroGraph.
Once you have your AllegroGraph license key, follow the instructions in kabob.app.git/allegrograph/build/config/user-env.sh.example to create a user-env.sh file with your AllegroGraph license key. Place the newly created user-env.sh file in the same directory as the user-env.sh.example file.
At this point, the KaBOB build is ready to proceed via a succession of scripts that call Docker commands. All scripts should be run from the base directory of the project:
cd kabob.app.git
The first step of the build downloads a collection of publicly available data sources and subsequently generates RDF versions of these sources that are grounded in classes defined in an extension of the Information Artifact Ontology. Optionally, the user can also provide the local path to some data sources which are not directly downloadable (often due to licensing restrictions).
Run: scripts/step1_rdf-gen.sh -k [KEY] -c [n] -d [DRUGBANK_XML_FILE] -p [PHARMGKB_RELATIONSHIPS_FILE]
where:
- [KEY] is a user-defined key to uniquely identify the KaBOB build. This key enables multiple KaBOB instances to be run in the same Docker environment. Example keys may be "development" or "production". Keys must not contain whitespace.
- [n] is the number of docker containers (1-5) that will be used to generate RDF. n should be <= the number of cores available on your machine.
- [DRUGBANK_XML_FILE] is the path to the DrugBank 'full database.xml' file on the local file system. The DrugBank 'full database.xml' file can be downloaded from here after creating an account and agreeing to the DrugBank license. This argument is optional. If the user prefers to exclude DrugBank from the KaBOB build then this argument can be excluded.
- [PHARMGKB_RELATIONSHIPS_FILE] is the path to the PharmGKB relationships file (relationships.tsv) on the local file system. Use of this file requires a PharmGKB license which can be obtained here. Note: This argument is optional. If the user prefers to exclude the PharmGKB relationships from the KaBOB build then this argument can be excluded.
This step may take >90 min depending on how many docker containers are used.
Run: scripts/ag-specific-scripts/step2_ag-setup.sh -k [KEY] where:
- [KEY] is the same user-defined key specified in Build Step 1 above that uniquely identifies the KaBOB build.
At this point, AllegroGraph should be running and its WebView UI should be visible at http://[HOST_URL]:[PORT], where [HOST_URL] is the URL for the machine hosting KaBOB. Access credentials and the [PORT] for logging into AllegroGraph can be found in the
user-env.shfile created earlier in the AllegroGraph Build Setup.
Run: scripts/ag-specific-scripts/step3_build-kabob.sh -k [KEY] -n [KB_NAME] where:
- [KEY] is the same user-defined key specified in Build Step 1 above that uniquely identifies the KaBOB build.
- [KB_NAME] is the name of the AllegroGraph repository that will be used to store this instance of KaBOB, e.g. 'kabob-human'
Building the human KaBOB instance should take ~100 minutes. If you would like to follow along via the agraph logs you can login to the agraph container using
docker exec -ti agraph bashand then view the agraph log output usingtail -f /tmp/agraph_load_check---supervisor-MKGnli.log(note the name of the log file may be slightly different)