Over 15 years ago, Sir Tim Berners Lee proclaimed the founding of an exciting new future involving intelligent agents operating over smarter data in order to perform complex tasks at the behest of their human controllers. At the heart of this vision lies an uneasy alliance between tedious formal knowledge representations and powerful analytics over big, but often messy data. Bio2RDF, our decade old open source project to create Linked Data for the life sciences, has weaved emergent Semantic Web technologies such as ontologies and Linked Data to generate FAIR - Findable, Accessible, Interoperable, and Reusable - data in the form of billions of machine accessible statements for use in downstream biomedical discovery. This revolution in data publication has been strengthened by action from global bioinformatics institutions such as the NCBI, NCBO, EBI, and DBCLS. Notably, NCBI's PubChem has successfully coupled large scale data integration with community-based standards to offer a remakable biochemical knowledge resource amenable to data hungry discovery tools. Yet, in the face of increasing pressure from researchers, funders, and publishers, will these approaches be sufficient for growing and maintaining a comprehensive knowledge graph that is inclusive of all biomedical research?

1. Building a network of interoperable and
independently produced linked and open
biomedical data
1
Michel Dumontier, Ph.D.
Associate Professor of Medicine (Biomedical Informatics)
Stanford University
@micheldumontier::ACS:23-08-16
An invited talk in support of the 2016 Herman Skolnik Awardees

2. My research aims to
develop computational
methods for biomedical
knowledge discovery
We develop tools and
methods to represent,
store, publish, integrate,
query, and reuse
biomedical data,
software, and ontologies
@micheldumontier::ACS:23-08-162

3. @micheldumontier::ACS:23-08-163
reuse needs to be
considered firmly in the
context of discovery and
reproducibility

4. 4 @micheldumontier::ACS:23-08-16
Most published research findings are false
- John Ioannidis, Stanford University

5. Reproducible discovery
1. Data Science Tools and Methods
– Infrastructure: To identify, annotate, link, integrate,
search for and query data and services
– Tools: To identify and uncover support for known or
novel associations
2. Community Standards
to contribute to and interrogate a massive, decentralized
network of interconnected data and software
@micheldumontier::ACS:23-08-165

6. @micheldumontier::ACS:23-08-166
FAIR: Findable, Accessible, Interoperable, Re-usable

7. FAIR: Findable, Accessible,
Interoperable, Re-usable
Findable
– Globally unique identifiers for datasets and the data they contain
– Rich set of descriptors to search and filter with
– Indexed and searchable
Accessible
– Identifiers can be used to retrieve representations using standard protocols
(e.g. HTTP)
– Metadata is always available.
Interoperable
– Data represented with formal knowledge representations
– Include links to other datasets/vocabularies
Reusable
– Licensing, Provenance, Community standards
@micheldumontier::ACS:23-08-167

8. The Semantic Web
is the new global web of knowledge
8 @micheldumontier::ACS:23-08-16
standards for publishing, sharing and querying
facts, expert knowledge and services
scalable approach for the discovery
of independently formulated
and distributed knowledge

9. Linked Data
offers a solid foundation for FAIR data
• Entities (people, proteins, pathways, etc) are
identified using globally unique identifiers (URIs)
• Entity descriptions are represented with a
standardized language (RDF)
• Data can be retrieved using a universal protocol
(HTTP)
• Entities (concepts, data, resources) can be linked
together to increase interoperability
@micheldumontier::ACS:23-08-169

10. @micheldumontier::ACS:23-08-16
Linked Data for the Life Sciences
10
Bio2RDF is an open source project to unify the
representation and interlinking of biological data using RDF.
chemicals/drugs/formulations,
genomes/genes/proteins, domains
Interactions, complexes & pathways
animal models and phenotypes
Disease, genetic markers, treatments
Terminologies & publications
• 11B+ interlinked statements from 35 biomedical
datasets and 400+ ontologies
• dataset description, provenance & statistics
• A growing interoperable ecosystem with the EBI,
NCBI, DBCLS, NCBO, OpenPHACTS, and
commercial tool providers

11. Bio2RDF
normalizes identifiers, formats, links, and access
11 @micheldumontier::ACS:23-08-16

12. @micheldumontier::ACS:23-08-1612

13. Bio2RDF shows how datasets are
connected together
@micheldumontier::ACS:23-08-1613

14. Queries can be federated across
private and public SPARQL databases
Get all protein catabolic processes (and more specific GO terms) in biomodels
SELECT ?go ?label count(distinct ?x)
WHERE {
service <http://bioportal.bio2rdf.org/sparql> {
?go rdfs:label ?label .
?go rdfs:subClassOf+ ?tgo
?tgo rdfs:label ?tlabel .
FILTER regex(?tlabel, "^protein catabolic process")
}
service <http://biomodels.bio2rdf.org/sparql> {
?x <http://bio2rdf.org/biopax_vocabulary:identical-to> ?go .
?x a <http://www.biopax.org/release/biopax-level3.owl#BiochemicalReaction> .
}
}
@micheldumontier::ACS:23-08-1614

15. Graph-like representation amenable
to finding mismatches and discovering new links
@micheldumontier::ACS:23-08-1615
W Hu, H Qiu, M Dumontier. Link Analysis of Life Science Linked Data.
International Semantic Web Conference (2) 2015: 446-462.

16. EbolaKB
Using Linked Data and Software
@micheldumontier::ACS:23-08-1616
Kamdar, Dumontier. An Ebola virus-centered knowledge base. Database. 2015 Jun 8;2015. doi: 10.1093/database/bav049.

17. @micheldumontier::ACS:23-08-1617
Network analysis and discovery
McCusker, McGuiness, Dumontier. In prep.

18. @micheldumontier::ACS:23-08-1618
Can we implement
an open version of
PREDICT using
Linked Data?
AUC 0.91 across all therapeutic indications
A. Chemical structure Similarity
B. Side Effect Similarity
C. Target Sequence Similarity
D. Target Functional Similarity
E. Network Distance
A. Phenotype Based
B. Text Extracted Concepts
Disease-disease similarityDrug-drug similarity

19. HyQue: Hypothesis Validation
• A platform for knowledge discovery that
uses data retrieval coupled with
automated reasoning to validate
scientific hypotheses
• Leverages semantic technologies to
provide access to linked data,
ontologies, and semantic web services
• Uses positive and negative findings,
captures provenance
• Weighs evidence according to context
• Used to find aging genes in worm,
assess cardiotoxicity of tyrosine kinase
inhibitors
HyQue: evaluating hypotheses using Semantic Web technologies. J Biomed Semantics. 2011 May 17;2 Suppl 2:S3.
Evaluating scientific hypotheses using the SPARQL Inferencing Notation. Extended Semantic Web Conference (ESWC 2012). Heraklion, Crete.
May 27-31, 2012. @micheldumontier::ACS:23-08-1619

20. What evidence might we gather?
• clinical: Are there cardiotoxic effects associated with the drug?
– Literature (studies) [curated db]
– Product labels (studies) [r3:sider]
– Clinical trials (studies) [r3:clinicaltrials]
– Adverse event reports [r2:pharmgkb/onesides]
– Electronic health records (observations)
• pre-clinical associations:
– genotype-phenotype (null/disease models) [r2:mgi, r2:sgd; r3:wormbase]
– in vitro assays (IC50) [r3:chembl]
– drug targets [r2:drugbank; r2:ctd; r3:stitch]
– drug-gene expression [r3:gxa]
– pathways [r2:kegg; r3:reactome]
– Drug-pathway, disease-pathway enrichments [aberrant pathways]
– Chemical properties [r2:pubchem; r2.drugbank]
– Toxicology [r1.toxkb/cebs]
@micheldumontier::ACS:23-08-1620

21. @micheldumontier::ACS:23-08-1621
HyQue

22. Beyond Bio2RDF
@micheldumontier::ACS:23-08-1622

23. Network of Linked Data (~2007)
@micheldumontier::ACS:23-08-1623

24. Expansion across domains
“Linking Open Data cloud diagram, by Richard Cyganiak and Anja Jentzsch. http://lod-cloud.net/”@micheldumontier::ACS:23-08-1624

25. A rapidly growing network of Linked Data
25 @micheldumontier::ACS:23-08-16Linking Open Data cloud diagram 2014, by Max Schmachtenberg, Christian Bizer, Anja Jentzsch and Richard Cyganiak. http://lod-cloud.net/"

26. @micheldumontier::ACS:23-08-1626

27. @micheldumontier::ACS:23-08-1627

28. @micheldumontier::ACS:23-08-1628

29. but the lack of coordination makes
Linked Open Data is chaotic and unwieldy
@micheldumontier::ACS:23-08-1629

30. There is no shortage
of vocabularies, ontologies and community-based
standards
@micheldumontier::ACS:23-08-1630

31. 68 168
@micheldumontier::ACS:23-08-1631

32. 3
2
metadatacenter.org
NIH COMMONS
@micheldumontier::ACS:23-08-16
Making it Easier, Possibly Even Pleasant, to Author
Interoperable Experimental Metadata

33. PubChem engaged the community to
reuse and extend existing vocabularies
@micheldumontier::ACS:23-08-1633

34. @micheldumontier::ACS:23-08-1634
Semanticscience Ontology (SIO)
An effective upper level ontology.
1500+ classes
207 object properties (inc. inverses)
1 datatype property

35. Chemical Information Ontology
(CHEMINF)
• Collaborative ontology
• Distinguishes algorithmic,
or procedural information
from declarative, or factual
information, and renders of
particular importance the
annotation of provenance
to calculated data.
@micheldumontier::ACS:23-08-1635

36. Where are we going?
• Large scale publishing on the web across
biomedical datatypes is possible on the web
• Hubs, such as NCBI and EBI now integrate data,
but there is need for global coordination on all
datatypes
• Standard Vocabularies must to be open, freely
accessible, and demonstrably reused
• Use of worldwide data integration formats (RDF)
and improved linking of data
• Easier to deploy toolkits for providing standards-
compliant linked data
@micheldumontier::ACS:23-08-1636

37. Linked Data Platform
Docker
• Data conversion scripts
• Query Editor
• Faceted Browser
• Relation Exploration
• API
• Data and data store
Model Organism Linked Data
MO-LD.org
37

38. In Summary
• We use semantic technologies such as ontologies
and linked data to make sense of and facilitate
access to biomedical data (FAIR)
• The intimate development and use of standards
by PubChem and others brings us closer to an
interoperability ideal
• Much more work is needed to support
(computational) discovery in a reproducible
manner.
@micheldumontier::ACS:23-08-1638

39. Acknowledgements
Dumontier Lab
• Amrapali Zaveri
• Mary Panahiazar
• Shima Dastgheib
• Sandeep Ayyar
• Remzi Celebi
• David Odgers
• Wei Hu
• Ruben Verborgh
• Leo Chepelev
• Alison Callahan
• Jose Miguel Toledo Cruz
• Tanya Hiebert
• Beatriz Lujan
+ many more
Collaborators
• Mark Musen
• Nigam Shah
• Robert Hoehndorf
• Janna Hastings
• Christoph Steinbeck
• Egon Willighagen
• Nico Adams
• Colin Batchelor
• David Wild
• Evan Bolton
• Gang Fu
+ many more
@micheldumontier::ACS:23-08-1639

40. dumontierlab.com
michel.dumontier@stanford.edu
Website: http://dumontierlab.com
Presentations: http://slideshare.com/micheldumontier
40 @micheldumontier::ACS:23-08-16