NUI Galway – UL Alliance First Annual ENGINEERING AND - ARAN ...

Improving discovery in Life Sciences and Health Care with Semantic Web
Technologies and Linked Data
Helena F. Deus and Jonas S. Almeida
Digital Enterprise Research Institute, National University of Ireland, Galway
Email: helena.deus@deri.org
Abstract
One of the most enticing outcomes of biological
exploration for the quantitative-minded researcher is to
identify the “mathematics of biology” through the study
of the patterns and interrelatedness of biological
entities. To move forward in that direction, many pieces
need to be set in place, from the availability of
biological data in forms that can be computationally
manipulated to the automated discovery of patterns that
would derive from integration of data in many and
diverse endpoints. A computational system where
researchers could securely deposit any type of data and
have it immediately analyzed, traversed, annotated and
merged with data deposited elsewhere is a dream not
yet achieved but one which could revolutionize
scientific discovery and, ultimately, help cure disease.
1. Introduction
“Science is organized knowledge”, said Immanuel
Kant. Science in the Information Age has indeed been
characterized by increasingly complex approaches to the
organization and sharing of scientific knowledge to
improve its discovery. A critical bottleneck in applying
knowledge engineering to inform and improve
biomedical discovery is the ability to align experimental
data acquisition with knowledge representation models.
Recent technological advances in genomics, proteomics
and other ‘omics’ sciences, have flooded microbiology
labs with unprecedented amounts of experimental data
and were at the heart of a paradigm shift for researchers
interested in its computational representation and
analysis.
2. Knowledge Continuum for Life Sciences
The deluge of data in biology marked the beginning
of the “industrialization of data production in Life
Sciences beyond a craft-based cottage industry” [1].
The successful exploit of this new wave of data
acquisition technologies created the need for multiple,
overlapping, sub-disciplines of biology to cope with the
increasing complexity of biological results. However,
this became an obstacle to discovery because the
answers to biological problems often span multiple
domain boundaries and rely on the existence of
knowledge continuums. Starting to address these
challenges requires agreement on optimal strategies for
publishing experimental biomedical data in
interoperable formats even before they are made
available to the community through peer-reviewed
publication.
129
3. Architecture
The work presented here corresponds to the
identification of the key requirements for supporting
publication of experimental biomedical results in the
Web of Data. With that purpose, the prototypical
application S3DB (Simple Sloppy Semantic Database)
was developed to fit the requirements of a Linked Data
Knowledge Organization System (KOS) for the Life
Sciences [3]. Reflecting our data-driven approach, the
technological requirements and advancements in the
S3DB prototype were iteratively devised through direct
interaction and validation by biomedical domain
experts, who interact with S3DB using graphical user
interfaces (GUI). Application developers access and
consume data in S3DB through a SPARQL application
programming interface (API) whereas the interaction
between multiple S3DB systems is ensured through a
domain specific language, S3QL, which operates on
S3DB’s indexing schema.
4. References
[1] C. Goble and R. Stevens, “State of the nation in data
integration for bioinformatics.,” Journal of biomedical
informatics, vol. 41, Oct. 2008, pp. 687-93.
[2] C. Brooksbank and J. Quackenbush, “Data standards: a
call to action.,” Omics : a journal of integrative biology,
vol. 10, Jan. 2006, pp. 94-9.
[3] H.F. Deus, R. Stanislaus, D.F. Veiga, C. Behrens, I.I.
Wistuba, J.D. Minna, H.R. Garner, S.G. Swisher, J.A.
Roth, A.M. Correa, B. Broom, K. Coombes, A. Chang,
L.H. Vogel, and J.S. Almeida, “A Semantic Web
management model for integrative biomedical
informatics.,” PloS one, vol. 3, Jan. 2008, p. e2946.