Deborah McGuiness
Udo Hahn
 Yves Lussier
 Trey Ideker

 


Keynote speakers


Deborah McGuiness

Rensselaer Polytechnic Institute (RPI)
Troy, NY, USA

Dr. Deborah McGuinness is the Tetherless World Senior Constellation Chair, Professor of Computer and Cognitive Science, and founding director of Rensselaer Polytechnic Institute’s Web Science Research Center. Deborah is a leading authority on the semantic web and has been working in knowledge representation and reasoning environments for over 25 years. Her primary research focuses on making smart systems understandable and usable by a broad range of people.  She leads active research efforts in explanation, trust, ontology environments, and provenance. Deborah is also known for semantic application environments, particularly for eScience frameworks such as the Semantic eScience Framework and demonstration portals including many in natural science and health informatics settings.  Deborah also founded McGuinness Associates – a small woman owned business - that consults on semantic applications in a wide range or areas with recent focus on health and environmental informatics, context-aware mobile computing, and next generation journalism.




back to top.


Udo Hahn

The Language and Information Engineering Lab
University of Jena, Jena, Germany

Udo Hahn received his Doctoral degree in Information Science (1987) from the University of Konstanz, Germany, and his Master’s degree in Linguistics (1980) from the University of Mainz. He is currently a Full Professor of Computational Linguistics and Language Technology at Friedrich-Schiller-Universität Jena (Germany) where he is the Head of the Jena University Language & Information Engineering (JULIE) Lab (http://www.julielab.de/) and also the Director of the Frege Centre for Structural Sciences (http://www.frege.uni-jena.de/). Formerly, he was affiliated as an Associate Professor for Language Informatics with the University of Freiburg (1990-2004) and as an Assistant Professor for Computer Science (Information & Database Systems) with the University of Passau (1987-1990). His main areas of interest include biomedical natural language processing (with emphasis on information extraction and text mining) as well as large-scale ontology design and ontology engineering for the life sciences. He has (co-)authored and (co-)edited 4 books, 9 proceedings, 45 journal articles, 25 contributions to edited volumes and 240 peer-reviewed proceedings papers; his current h-index is 36. In recognition of his scientific achievements he has received the IBM University Partnership Faculty Award in 2000, the IBM UIMA Innovation Award for Faculty in 2007 and the IBM Unstructured Information Analytics Award in 2008.



Information Extraction in the Full-Text Age – Can Current Ontologies Cope with Elaborate Text Structure, and why should they?


With the ever-increasing availability of full texts (original journal articles, clinical notes, patents, etc.) entirely new types of information sources can be mined by computational text analytic techniques. A full text's informational richness exceeds that of document surrogates (such as abstracts and titles, e.g. taken from Medline). that we have mostly been working with in the past decades within the life sciences community) by several orders of magnitude and, thus, might be a real gold mine to harvest in depth. This is the good news.
Yet, there is no free lunch. Full texts are a different and much more complex text genre compared with abstracts. Linguistically, the high volume of single information pieces contained in full texts gets organized by several text structuring mechanisms which have immediate and deeply rooted links to the underlying domain knowledge. In this talk, I shall discuss several crucial types of linguistic phenomena which commonly occur in full texts at the linguistic expression level. Prime among them are various types of anaphora, i.e. linguistic mechanisms that link discourse units (the theme(s) being talked about) referentially such as illustrated in the following sequence of clause skeletons “Interleukin-7 … This protein [= Interleukin-7] … It [= Interleukin-7] …” or establish non-referential links as illustrated by “the cell … Inside we find the cytoplasmic region [of the cell] which …” I will show how such phenomena (and others) establish coherent text structure by systematically addressing conceptual correlates at the ontological level – basically, taxonomic [Interleukin-7 –is-a – protein] and partonomic [cell – has-part – cytoplasmic region] concept structures and corresponding reasoning mechanisms.
We shall conclude that full-text-based information extraction is likely to fail, by the unwarranted synthesis of invalid, incomplete and incoherent knowledge bases, unless adequate support from ontologies is provided to cope with the new text structure challenges implied by full text analysis. This relates to both sides of the coin – the computational infrastructure provided and domain-specific contents captured in concrete ontologies which should support such taxonomic and partonomic relations. Is this the bad news?

back to top.


Yves Lussier

University of Illinois College of Medicine at Chicago
Chicago, IL, USA

Dr. Lussier is the Assistant Vice President for Health Affairs and Clinical Research Information Officer for the University of Illinois Hospital and Health Science System. He also serves as a Professor of Medicine and Bioengineering. He holds medical and engineering degrees from the University of Sherbrooke (Canada) and completed a post-doctoral training in the Dept. of Biomedical Informatics at Columbia University under Drs. James J. Cimino and Carol Friedman. He practiced as a general internist from 1992 until his induction in the American College of Medical Informatics as a Fellow in 2005. His research focuses on hypothesis-driven translational bioinformatics methods to accurately personalize the understanding, prediction and treatments of diseases. To date, his team has developed solutions powered with ontologies: (i) the 1st tablet-based electronic medical records anchored on the 1st commercial ontology (Purkinje.com, 1991-; 65,000 terms, 7 semantic types), (ii) the 1st ontology-anchored clinical event monitor of the Columbia University New York Presbyterian Hospital (Vigilens, IBM awards, 6x10 6 patients, operations x 2002), (iii) bioinformatically-predicted and network-based microRNA tumor suppressor confirmed in vitro and in vivo (ISCB award, PLoS Comput Biol, 2010 6(4): e1000730), (iv) network-based rescue and repositioning of a targeted therapy validated in vitro and in vivo (Multicenter randomized phase II clinical trial; NIH 3UL1RR024999-03S3). Lussier has served on over a dozen of governance, scientific and editorial boards (e.g. NLM, NIGMS, NASA, IBM Center for Computational Medicine, College of American Pathologists, Cerner, NIH/NLM, NSF, JAMIA, BMC Bioinformatics, etc.). He cumulates over 130 journal publications and $115M of grants as principal or co-investigator in addition to mentoring more than 35 graduate and postgraduate students and 11 junior Faculty members.



Personalized therapeutics powered by ontology-transforms


Translating genomic and molecular bioinformatics discoveries into clinical practice remains challenging. This presentation describes “ontology transforms,” which we define as the often-overlooked model transformations that can be performed over ontology-encoded datasets. For example, hierarchies of ontologies allow for scalar transformations and dimensionality reduction. Genesets enrichment (GSE) is a statistical hierarchical-transform conducted one genomic datasets annotated with ontologies and contrast from the deterministic hierarchical classification. GSE transform the signal from the gene level to that of an intermediate phenotype (mesophenotype). Deterministic classification as well as statistically derived “mesophenotypes” can be represented in the framework of Blois’s Theory of Biomedical Informatics as emerging properties. We will then show ontology-transforms in action with several scientific examples. For example, we hypothesized a network-based mechanism of resistance of head and neck cancers. We identified a FDA-approved drug that may sensitize resistant cancer cells by inhibiting the bioinformatically-identified aberrant molecular signals sustaining EGFR pathway activity. This network-targeting therapy, later validated in vitro/in vivo, spawned the development of our NIH Phase II clinical trial. We will also provide details of a recent study aimed at associating protein domain mechanisms with Mendelian, complex and unique personal genetics. We conclude with some broader considerations on the utility of ontology transforms and mesophenotypes to predict clinical outcome.

back to top.

Trey Ideker

School of Medicine, UC San Diego
San Diego, CA, USA

Dr. Ideker is Chief of Genetics at the UCSD School of Medicine. He also serves as Professor of Bioengineering, Adjunct Professor of Computer Science and Member of the Moores UCSD Cancer Center. Ideker received Bachelor’s and Master’s degrees from MIT in Electrical Engineering and Computer Science and his Ph.D. from the University of Washington in Molecular Biology under the supervision of Dr. Leroy Hood. He is a pioneer in using genome-scale measurements to construct network models of cellular processes and disease. His recent research activities include assembly of networks governing the response to DNA damage, development of software for protein network cross-species comparisons, and network-based diagnosis of disease. Ideker serves on the Editorial Boards for Bioinformatics and PLoS Computational Biology, Board of Directors for US-HUPO and the Cytoscape Consortium, and is a regular consultant for companies such as Monsanto, Genstruct, and Mendel Biotechnology. He was named one of the Top 10 Innovators of 2006 by Technology Review magazine and the 2009 Overton Prize recipient from the International Society for Computational Biology. His work has been featured in news outlets such as The Scientist, the San Diego Union Tribune, and Forbes magazine. 



A gene ontology constructed from molecular networks


Ontologies have proven very useful for capturing knowledge as a hierarchy of terms and their interrelationships. In biology a major challenge has been to construct ontologies of gene function given incomplete biological knowledge and inconsistencies in how this knowledge is manually curated. Here we show that large networks of gene and protein interactions in Saccharomyces cerevisiae can be used to infer an ontology whose coverage and power are equivalent to those of the manually curated Gene Ontology (GO). The network-extracted ontology (NeXO) contains 4,123 biological terms and 5,766 term-term relations, capturing 58% of known cellular components. We also explore robust NeXO terms and term relations that were initially not cataloged in GO, a number of which have now been added based on our analysis. Using quantitative genetic interaction profiling and chemogenomics, we find further support for many of the uncharacterized terms identified by NeXO, including multisubunit structures related to protein trafficking or mitochondrial function. This work enables a shift from using ontologies to evaluate data to using data to construct and evaluate ontologies.

back to top.