oaklib 0.2.2

Ontology Access Kit: Python library for common ontology operations over a variety of backends

Ontology Access Kit

Python lib for common ontology operations over a variety of backends.

OAK provides a collection of interfaces for various ontology operations, including:

• look up basic features of an ontology element, such as its label, definition, relationships, or aliases
• search an ontology for a term
• validate an ontology
• modify or delete terms
• generate and visualize subgraphs
• identify lexical matches and export as SSSOM mapping tables
• perform more advanced operations, such as graph traversal, OWL axiom processing, or text annotation

These interfaces are separated from any particular backend. This means the same API can be used regardless of whether the ontology:

• is served by a remote API such as OLS or BioPortal
• is present locally on the filesystem in owl, obo, obojson, or sqlite formats
• is to be downloaded from a remote repository such as the OBO library
• is queried from a remote database, including SPARQL endpoints (Ontobee/Ubergraph), A SQL database, a Solr/ES endpoint

Documentation:

• incatools.github.io/ontology-access-kit
• Inaugural OAK Workshop slides
• OBO Tools Session slides

Contributing

See the contribution guidelines at CONTRIBUTING.md. All contributors are expected to uphold our Code of Conduct.

Usage

from oaklib import OntologyResource

ontology_resource = OntologyResource(slug='tests/input/go-nucleus.db', local=True)
ontology_interface = ontology_resource.materialize("sql")
# can also pass an implementation class explicitly instead of a string.

for curie in ontology_interface.basic_search("cell"):
    print(f'{curie} ! {ontology_interface.label(curie)}')
    for rel, fillers in ontology_interface.outgoing_relationship_map(curie).items():
        print(f'  RELATION: {rel} ! {ontology_interface.label(rel)}')
        for filler in fillers:
            print(f'     * {filler} ! {ontology_interface.label(filler)}')

For more examples, see

• demo notebook

Command Line

Documentation here is incomplete.

See CLI docs

Search

Use the pronto backend to fetch and parse an ontology from the OBO library, then use the search command

runoak -i obolibrary:pato.obo search osmol 

Returns:

PATO:0001655 ! osmolarity
PATO:0001656 ! decreased osmolarity
PATO:0001657 ! increased osmolarity
PATO:0002027 ! osmolality
PATO:0002028 ! decreased osmolality
PATO:0002029 ! increased osmolality
PATO:0045034 ! normal osmolality
PATO:0045035 ! normal osmolarity

QC and Validation

Perform validation on PR using sqlite/rdftab instance:

runoak -i sqlite:../semantic-sql/db/pr.db validate

List all terms

List all terms obolibrary has for mondo

runoak -i obolibrary:mondo.obo terms 

Lexical index

Make a lexical index of all terms in Mondo:

runoak  -i obolibrary:mondo.obo lexmatch -L mondo.index.yaml

Search

Searching over OBO using ontobee:

runoak  -i ontobee: search tentacle

yields:

http://purl.obolibrary.org/obo/CEPH_0000256 ! tentacle
http://purl.obolibrary.org/obo/CEPH_0000257 ! tentacle absence
http://purl.obolibrary.org/obo/CEPH_0000258 ! tentacle pad
...

Searching over a broader set of ontologies in bioportal (requires API KEY) (https://www.bioontology.org/wiki/BioPortal_Help#Getting_an_API_key)

runoak set-apikey bioportal YOUR-KEY-HERE
runoak  -i bioportal: search tentacle

yields:

BTO:0001357 ! tentacle
http://purl.jp/bio/4/id/200906071014668510 ! tentacle
CEPH:0000256 ! tentacle
http://www.projecthalo.com/aura#Tentacle ! Tentacle
CEPH:0000256 ! tentacle
...

Alternatively, you can add "BIOPORTAL_API_KEY" to your environment variables.

Searching over more limited set of ontologies in Ubergraph:

runoak -v -i ubergraph: search tentacle

yields

UBERON:0013206 ! nasal tentacle

Annotating Texts

runoak  -i bioportal: annotate neuron from CA4 region of hippocampus of mouse

yields:

object_id: CL:0000540
object_label: neuron
object_source: https://data.bioontology.org/ontologies/NIFDYS
match_type: PREF
subject_start: 1
subject_end: 6
subject_label: NEURON

object_id: http://www.co-ode.org/ontologies/galen#Neuron
object_label: Neuron
object_source: https://data.bioontology.org/ontologies/GALEN
match_type: PREF
subject_start: 1
subject_end: 6
subject_label: NEURON

...

Mapping

Create a SSSOM mapping file for a set of ontologies:

robot merge -I http://purl.obolibrary.org/obo/hp.owl -I http://purl.obolibrary.org/obo/mp.owl convert --check false -o hp-mp.obo
runoak lexmatch -i hp-mp.obo -o hp-mp.sssom.tsv

Visualization of ancestor graphs

Use the sqlite backend to visualize graph up from 'vacuole' using test ontology sqlite:

runoak -i sqlite:tests/input/go-nucleus.db  viz GO:0005773

Same using ubergraph, restricting to is-a and part-of

runoak -i ubergraph:  viz GO:0005773 -p i,BFO:0000050

Same using pronto, fetching ontology from obolibrary

runoak -i obolibrary:go.obo  viz GO:0005773

Configuration

OAK uses pystow for caching. By default, this goes inside ~/.data/, but can be configured following these instructions.

Developer notes

Local project setup

Prerequisites:

1. Python 3.9+
2. Poetry

Setup steps:

git clone https://github.com/INCATools/ontology-access-kit.git
cd ontology-access-kit
poetry install

Testing locally:

poetry run python -m unittest discover

Code quality locally:

poetry run tox

Potential Refactoring

Currently all implementations exist in this repo/module, this results in a lot of dependencies

One possibility is to split out each implementation into its own repo and use a plugin architecture