BioPortal as (the only functional)

OOR SandBox(so far)

Natasha Noy, Michael Dorf

Stanford University

BioPortal: A Community-Based Ontology Repository

http://bioportal.bioontology.org

The National Center for Biomedical Ontology

• One of the five National Centers for Biomedical Computing launched by NIH

• Collaboration of Stanford, Mayo, Buffalo, Victoria, Medical College of Wisconsin, Washington University, John Hopkins

• Primary goal is to make ontologies accessible and usable• Research will develop technologies for ontology

dissemination, indexing, alignment, and peer review

Key Technology: BioPortal

• Web accessible repository of ontologies for the biomedical community:– http://bioportal.bioontology.org

• Online support for ontology– Peer review– Notes (comments and discussion)– Versioning– Mapping– Search– Resources

The BioPortal Technology

• BioPortal is a library of biomedical ontologies• All BioPortal data is accessible through REST

services– BioPortal user interface accesses the repository

through REST services as well

• The BioPortal technology is domain-independent– There are installations of BioPortal for libraries in

other domains– BioPortal code is open-source– Technology stack includes mySQL, J2EE, Spring,

Hibernate, Ruby-on-Rails

OOR and BioPortal

• The OOR sandbox currently runs the BioPortal code

• The assumption is that it will diverge• Some of the OOR requirements contribute to

the setting of directions for BioPortal• But BioPortal has its user base that primarily

drives the requirements (and of course, the funders)

OOR Requirements (not addressed -- yet) in BioPortal

• Persistence– No triplestore-based storage (planned)

• Rules– Unlikely in the next year

• Policies– Will have some support (private areas of

BioPortal)

Representation Languages• Not trivial to add support for Common Logic

and other expressive formalisms• A simple data model drives the user interface

– Hence, you cannot easily adapt what will show up in the class form

• You can add an ontology in CL, for example, and make it available for download, but not for browsing, etc.– Perhaps not that useful

Ontology Services• Syntax validation and consistency checking

– May add, but many biomedical ontologies don’t have enough expressive power to check consistency

• Editing and creation of ontologies– Not in BioPortal, but supported by WebProtégé (

http://protegewiki.stanford.edu/index.php/WebProtege)

– We are about to start the work on integrating BioPortal and WebProtégé to enable a tight link between editing and library functions

Auditing• We have rudimentary auditing, quality

control of contributed ontologies

• We hope to get more community involvement in this task– The infrastructure for reviewing and

ranking is mostly set up– Need users to contribute to these areas of

BioPortal

Discovery

• Metadata search is a weak point– Will be addressed in the next few months– Will enable search of comments,

mappings, reviews

Metadata Representation in BioPortal

• As of September 2009, BioPortal uses ontology-based representation of metadata

• The following are part of the metadata and are represented as instances in the BioPortal metadata ontology:– Ontology details– notes and comments (soon)– mappings (soon)– reviews (soon)– ontology views

Metadata in BioPortal

Knowledge Domain

• BioPortal itself is a repository of biomedical ontologies

• However, its technology is domain-independent and will remain so

Short- to Mid-Term Plans

• Migration of the rest of the metadata to ontology-based representation

• Structured notes to collect term proposals• Integration of BioPortal and WebProtégé• Ontology evaluation metrics• Access controls; private areas of BioPortal• myBioPortal

Long-Term Plans

• Federation

• Triplestore as storage mechanism

• Generation of custom-tailored values sets and dictionaries

• ???