The Ontology for Biomedical Investigations (OBI) helps you communicate clearly about scientific investigations by defining more than 2500 terms for assays, devices, objectives, and more.
This is the developer repository for OBI. You can download the most up-to-date OBI products here and learn more about OBI through our documentation.
Our ontology terms come in three groups. Depending on what type of term you want to edit or add, you have to go through different routes:
-
external terms (from other ontologies): We use OntoFox for imports. Edit the corresponding
src/ontology/OntoFox_inputs/
file. -
template terms: We use ROBOT templates to convert spreadsheets to OWL. Edit the relevant
src/ontology/templates/
file:obsolete.tsv
for obsolete termsassays.tsv
for general assaysepitope-assays.tsv
specifically for immune epitope assaysvalue-specifications.tsv
specimens.tsv
for specimensmedical-history.tsv
for medical history classifications and related selection criteriastudy-designs.tsv
for study designsdata-sets.tsv
for data sets
-
other terms: Edit
src/ontology/obi-edit.owl
in Protege.
See below for a full list of files, build instructions, and instructions on using Git and GitHub for OBI.
If you wish to import terms from an ontology for which OBI does not currently have an OntoFox import file (src/ontology/OntoFox_inputs/
), follow these steps:
- Write the import file for the new ontology (see existing import files for reference).
- Update
catalog-v001.xml
to list the new import file like it lists the existing import files. - Add an import statement to
obi-edit.owl
. - Create a blank file for the output module in
src/ontology/OntoFox_outputs/
, e.g., to make phony output for foo.owl imports:
touch src/ontology/OntoFox_outputs/foo_imports.owl
- Build the module & obi.owl.
make imports
make obi.owl
If you wish to edit a template or templates in Excel, rather than copy & pasting the template, we ask that you follow this workflow to preserve quoting. Going back and forth with Excel can cause some unintentional changes to double quotes within templates.
First, install the python requirements:
python3 -m pip install -r requirements.txt
Then, make the Excel sheet. In your local OBI git directory, run the following command to create a file called obi.xlsx
:
make obi.xlsx
Next, open obi.xlsx
in Excel (or whatever editor you prefer). This spreadsheet contains a tab for each OBI template (e.g., "study-design", "assays", etc.). Find the tab that corresponds to the template you need to edit, make your changes, and save the Excel spreadsheet to the same location (obi.xlsx
). Finally, run the following to update the TSV versions of the templates:
make update-tsv
This will convert the tabs in obi.xlsx
back to TSVs and overwrite the existing TSVs in the src/ontology/templates/
directory with your changes. Review your changes (git diff
) and make your pull request.
To find where a term lives, you can use src/scripts/locate.py
.
Then you can run the script to find terms by ID or label by passing them as a space-separated list, for example:
src/scripts/locate.py OBI:0000070 CHMO:0000087 GO:0000785
Labels should be enclosed in double quotes:
src/scripts/locate.py "assay" "fluorescence microscopy" "chromatin"
The OBI repo includes some Python scripts to help developers efficiently edit both obi-edit.owl
and OBI templates. These scripts require Python 3.
The scripts also require that you first to build two databases (one from OBI edit, and one from a merged version of OBI):
make obi-dbs
Since labels are used in templates, manually finding and replacing all usages of a term can be difficult and tedious. Instead, you can use src/scripts/relabel.py
to automatically update a term's label and its usages.
You can run this by passing the term you want to update and the new label:
src/scripts/relabel.py CHMO:0000087 "microscopy with fluorescence"
You can also pass the old label as the first argument, as long as it is enclosed in double quotes:
src/scripts/relabel.py "fluorescence microscopy" "microscopy with fluorescence"
Make sure to commit all changed files to ensure that all usages are updated.
README.md
this overview documentobi.owl
the latest release of OBIMakefile
scripts for building OBIviews/
various specialized views of OBIobi.obo
the latest release of OBI in.obo
file formatobi_core.owl
the latest release of OBI Core: ~100 key terms
src/
ontology/
source files for OBIobi-edit.owl
the main OBI OWL filecore.txt
the list of OBI Core termsexternal-byhand.owl
some custom imports from other ontologiescatalog-v001.xml
an artisinal list of OWL import overridestemplates/
ROBOT template files for various branches of OBImodules/
the results of the ROBOT templatesOntoFox_inputs/
OntoFox configuration files for importing from other ontologiesOntoFox_outputs/
OntoFox result files
sparql/
SPARQL queries for building and validating OBIscripts/
utility scriptsviews/
configuration for views
The Makefile
contains scripts for building OBI. On macOS or Linux, you should just be able to run make
or one of the specific tasks below. On Windows consider using some sort of Linux virtual machine such as Docker or Vagrant. Most results will be in the build/
directory. If you have trouble, contact James.
make test
merge and run SPARQL tests (this is run on every push to GitHub)make sort
sort templates, and fix quoting and line endingsmake imports
update OntoFox importsmake modules
update ROBOT templatesmake obi.owl
build the release file; reasoning can take about 10 minutesmake views
update ROBOT templatesmake all
prepare for a release, runsimports
,modules
,test
,obi.owl
, andviews
make build/obi_merged.owl
mergeobi-edit.owl
into a single file, don't reasonmake clean
remove temporary files
We use git and GitHub to develop OBI. There's a lot of good documentation on both:
Before you can start developing with OBI, you will need to do some initial setup:
-
sign up for a GitHub account
-
install the Git command line tool, the GitHub Desktop app, or another Git client of your choosing
-
if you're using macOS and Excel, set up a pre-commit hook (see below for details):
ln -s ../../src/scripts/check-line-endings.sh .git/hooks/pre-commit
Changes should be made in manageable pieces, e.g. add one term or edit a few related terms. Most changes should correspond to a single issue on the tracker.
Start from a local copy of the master
branch of the OBI repository. Make sure your local copy is up-to-date. Make your changes on a new branch. Please use the OBI Term ID Reservations sheet to manage new IDs.
When you're ready, push your branch to the OBI repository and make a Pull Request (PR) on the GitHub website. Your PR is a request to merge your branch back into master
. Your PR will be tested, discussed, adjusted if necessary, then merged. Then the cycle can repeat for the next change that you or another developer will make.
These are the steps with their CLI commands. When using a GUI application the steps will be the same.
git fetch
make sure your local copy is up-to-dategit checkout master
start on themaster
branchgit checkout -b your-branch-name
create a new branch named for the change you're making- make your changes
make sort
sort and normalize tables, for cleaner diffsgit status
andgit diff
inspect your changesgit add --update src/
add all updated files in thesrc/
directory to staginggit commit --message "Description, issue #123"
commit staged changes with a message; it's good to include an issue numbergit push --set-upstream origin your-branch-name
push your commit to GitHub- open https://github.com/obi-ontology/obi in your browser and click the "Make Pull Request" button
Your Pull Request will be automatically tested. If there are problems, we will update your branch. When all tests have passed, your PR will be reviewed by OBI developers. When that review is complete, a senior OBI developer will merge the PR. Rinse and repeat!
The easiest way to edit our src/ontology/template/
files is with Excel. Unfortunately Excel has some idiosyncratic rules for quoting cell values, and on macOS uses old line endings. Both these things make our diffs messy and confusing.
For clean diffs, we also like to keep out templates sorted by ID. The make sort
command will fix line endings and sorting by running all the templates through a Python script.