Accepted Posters

Category 'Y'- Text Mining'
Poster Y01
Gaining biological insights from text-mined facts through automated reasoning
Luis Tari- Hoffmann-La Roche
Chitta Baral (Arizona State University, Computer Science and Engineering); Saadat Anwar (Arizona State University, Computer Science and Engineering); Shanshan Liang (Arizona State University, Computer Science and Engineering); James Cai (Hoffmann-La Roche, Pharma Research Scientific Informatics);
Short Abstract: We present a solution that goes beyond existing text mining techniques to address deep biological questions such as pathway synthesis. Our approach involves the integration of text mining and automated reasoning techniques. We illustrate our approach with the synthesis of pharmacokinetic pathways and the identification of drug-drug interactions.
Long Abstract:Click Here

Poster Y02
Towards efficient search tools for biomedical databases: Characterizing user search habits and recognizing their information needs
Rezarta Islamaj Dogan- National Center for Biotechnology Information
Zhiyong Lu (National Center for Biotechnology Information, CBB); Craig Murray (National Center for Biotechnology Information, CBB); Aurelie Neveol (National Center for Biotechnology Information, CBB);
Short Abstract: Log analysis is a useful way to understand user needs and their search habits for improving information retrieval systems. Our analysis of PubMed logs data, consisting of millions of user queries and retrievals, provides useful insights for understanding the unique aspects of PubMed searches and PubMed users' perspective.
Long Abstract:Click Here

Poster Y03
Predicting chromosomal alterations from phenotypes based on scientific literature
Anika Oellrich- European Bioinformatics Institute
Dietrich Rebholz-Schuhmann (European Bioinformatics Institute, Text mining);
Short Abstract: Copy Number Variations (CNVs) cause specific phenotypes which characterise diseases and syndromes. We generate a mapping between phenotypic features and chromosomal alterations from the literature and then predict CNVs of patients whose phenotypes are similar to our phenotype profiles. Predictions are benchmarked against the DECIPHER database.
Long Abstract:Click Here

Poster Y04
Text Mining for Protein Function Prediction: Detection of Active Residues in Full-text publications
Karin Verspoor- U Colorado Denver
César Mejia Muñoz (University of Colorado Denver, );
Short Abstract: We describe the development of patterns for the recognition of mentions of specific amino acid residues in full-text journal publications, and their evaluation on a newly developed corpus of literature associated with known active sites in proteins.
Long Abstract:Click Here

Poster Y05
Bio-Jigsaw: A Biological Application of the Jigsaw System.
Hannah Tipney- University of Colorado, Denver
Carsten Görg (Georgia Institute of Technology, Atlanta, GA , School of Interactive Computing); William A. Baumgartner Jr. (University of Colorado, Denver, Center for Computational Pharmacology, School of Medicine); John Stasko (Georgia Institute of Technology, Atlanta, GA , School of Interactive Computing); Lawrence Hunter (University of Colorado, Denver, Center for Computational Pharmacology, School of Medicine);
Short Abstract: We introduce Bio-Jigsaw, a visual analytics system that behaves as a visual index over collections of biomedically important documents. We demonstrate Bio-Jigsaw's ability to support biologists in investigating and understanding connections between biological entities by its application to over 1500 primary papers returned from a "Breast cancer" PubMed query.
Long Abstract:Click Here

Poster Y06
Improving Biomedical Text Mining by using Existing Background Knowledge
Kevin Livingston- University of Colorado Denver
Karin Verspoor (University of Colorado Denver, Pharmacology); Lawrence Hunter (University of Colorado Denver, Pharmacology);
Short Abstract: Extending large bodies of knowledge is necessary for next generation genome-scale analysis of systems biology data. Most text mining methods are unidirectional in their interaction with existing knowledge: they only produce it. OpenDMAP is an approach that also leverages existing knowledge to better understand the text it is processing.
Long Abstract:Click Here

Poster Y07
Mining host-pathogen interactions from biomedical literature
Thanh Thieu- University of Missouri
Sneha Joshi (University of Missouri, Informatics Institute); Samantha Warren (University of Missouri, Undergraduate Program, Department of Mathematics); Dmitry Korkin (University of Missouri, Informatics Institute and Department of Computer Science);
Short Abstract: We present two literature mining approaches to automatically detect if a title or abstract of an article contain host-pathogen interaction data and extract the information about organisms and protein involved in the interaction. The first approach employs feature-based supervised learning and the second is based on natural language processing.
Long Abstract:Click Here

Poster Y08
Reflect: improving how biological research is communicated
Seán O'Donoghue- EMBL
Lars Jensen (University of Copenhagen, NNF Center for Protein Research); Evangelos Pafilis (EMBL, Structural and Computational Biology); Heiko Horn (University of Copenhagen, NNF Center for Protein Research); Michael Kuhn (Technische Universität Dresden, BIOTEC); Reinhard Schneider (EMBL, Structural and Computational Biology); Sven Haag (EMBL, Structural and Computational Biology);
Short Abstract: Reflect uses the Web 2.0 concept of augmented browsing: a single click highlights genes, proteins, and chemical names in any web page within seconds; clicking a highlighted name opens a popup showing key biochemical properties. Reflect can be installed as a plug-in to Firefox or Internet Explorer, or used via an API. Reflect won the Elsevier Grand Challenge, a contest for systems that improve how scientific information is communicated. Following the initial publication of Reflect in June 2009, usage has grown rapidly: over 35,000 scientists have installed Reflect, and over 4,000 web-pages are processed daily. Content providers are also using Reflect, e.g., Cell recently launched a pilot, ‘Reflecting’ three issues (Vol. 139 Nr. 4-6). Reflect was also featured in Nature (Vol, 463, 416-418 (2010)). In this talk we introduce Reflect, present upcoming features, and discuss our ideas about the future of such tools. Reflect is freely available at
Long Abstract:Click Here

Poster Y09
A test suite for ontology concept recognition systems: The Gene Ontology
K. Bretonnel Cohen- U. Colorado School of Medicine
Christophe Roeder (U. Colorado School of Medicine, Center for Computational Pharmacology); William A. Baumgartner Jr. (U. Colorado School of Medicine, Center for Computational Pharmacology); Lawrence E. Hunter (U. Colorado School of Medicine, Center for Computational Pharmacology); Karin Verspoor (U. Colorado School of Medicine, Center for Computational Pharmacology);
Short Abstract: A test suite was constructed for ontology concept recognition systems, focussing on the Gene Ontology. It was constructed using the methodology of structured testing. It was applied to an ontology concept recognition system and found to identify performance errors in it. General principles of test suite construction are also discussed.
Long Abstract:Click Here

Poster Y10
How to have a BLAST with your Desktop PC
Seth Mendelson- Novartis
No additional authors
Short Abstract: Patent analysts are periodically asked to extract a sequence from a patent and align it to a reference sequence. Three ways to decrease the burden of repetitive tasks will be shown: public patent sequence sources; search and replace commands used for cleanup of fetched sequence; and incorporation of some global commands into a macro.
Long Abstract:Click Here

Poster Y11
Machine Learning for Identifying Abbreviation Definitions
Lana Yeganova- National Institutes of Health
Donald Comeau (NIH, CBB); W. John Wilbur (NIH, CBB);
Short Abstract: Detecting and understanding abbreviations is an important component of text analysis tools, which are essential with exponential growth of electronic literature. In this study, we develop a machine learning based algorithm for abbreviation full form identification in text. Our preliminary results are comparable to the best rule-based approach.
Long Abstract:Click Here

Poster Y12
Building the Scientific Knowledge Mine (SciKnowMine ): a community-driven framework for text mining tools in direct service to biocuration
Cartic Ramakrishnan- USC - Information Sciences Institution
William Baumgartner (University of Colorado Denver, Center for Computational Pharmacology); Judith Blake (The Jackson Laboratory, MGI); Gully Burns (USC-Information Sciences Institute, Intelligent Systems Division); Kevin Cohen (University of Colorado Denver, Center for Computational Pharmacology); Harold Drabkin (The Jackson Laboratory, MGI); Janan Eppig (The Jackson Laboratory, MGI); Eduard Hovy (USC-Information Sciences Institute, Intelligent Systems Division); Chun-Nan Hsu (USC-Information Sciences Institute, Intelligent Systems Division); Lawrence Hunter (University of Colorado Denver, Center for Computational Pharmacology); Tommy Ingulfsen (USC-Information Sciences Institute, Intelligent Systems Division); Hiroaki Onda (The Jackson Laboratory, MGI); Sandeep Pokkunuri (University of Utah, Computer Science); Ellen Riloff (University of Utah, Computer Science); Christophe Roeder (University of Colorado Denver, Center for Computational Pharmacology); Karin Verspoor ( University of Colorado Denver, Center for Computational Pharmacology);
Short Abstract: The challenge of delivering effective computational support for curation of large-scale biomedical databases is still unsolved. In this paper we describe the SciKnowMine project aimed at providing knowledge engineering enhancements of existing biocuration systems via large-scale text processing pipelines bringing together multiple NLP tools developed using the UIMA framework.
Long Abstract:Click Here

Poster Y13
Scaling Text Mining to One Million Documents
Christophe Roeder- University of Colorado, School of Medicine
Karin Verspoor (University of Colorado School of Medicine, Pharmacology);
Short Abstract: Applying text mining to a large document collection demands more resources than the lab PC can provide. Preparing for such a task requires an understanding of the demands of the text mining software and capabilities of supporting hardware and software. We describe efforts to scale a large text mining task.
Long Abstract:Click Here

Poster Y14
CALBC Challenge I: first results
Dietrich Rebholz-Schuhmann- European Bioinformatics Institute
Amtonio Jimeno Yepes (European Bioinformatics Institute, Research); Chen Li (European Bioinformatics Institute, Research); Erik van Mulligen (Erasmus Medical Center, Medical Informatics); Jan Kors (Erasmus Medical Center, Medical Informatics); Ning Kang (Erasmus Medical Center, Medical Informatics); Katrin Tomanek (Friedrich-Schiller Universitaet, Julie Lab); Kerstin Hornbostel (Friedrich-Schiller Universitaet, Julie Lab); Udo Hahn (Friedrich-Schiller Universitaet, Julie Lab);
Short Abstract: CALBC aims at creating a large annotated corpus (on the order of 150,000 documents) with about 5 to 10 different semantic types (genes, diseases) whose annotation is carried out automatically. The CALBC challenge I has not been terminated (22 submissions) and first results are available.
Long Abstract:Click Here

Poster Y15
Considering alternative views when modeling cancer metastasis.
Anna Divoli- The University of Chicago
Andrey Rzhetsky (The University of Chicago, Department of Medicine, Department of Human Genetics, Institute for Genomic and Systems Biology, Computation Institute); Eneida Mendonca (The University of Chicago, Department of Pediatrics and Computation Institute);
Short Abstract: We interviewed 28 experts in cancer metastasis on their beliefs regarding ongoing metastasis-related research and analyzed their responses. Our data show a wide range of understandings as to how the process takes place. We believe this range should be reflected in computational models in order to produce more accurate simulations.
Long Abstract:Click Here

Poster Y16
MspecLINE: Bridging knowledge of human disease with the proteome
Sarah Killcoyne- Institute for Systems Biology
Sarah Killcoyne (Institute for Systems Biology, Shmulevich Lab, Computational Biology); John Boyle (Institute for Systems Biology, Shmulevich Lab, Computational Biology);
Short Abstract: Public proteomics databases such as PeptideAtlas contain peptides and proteins identi?ed in mass
spectrometry experiments. However, these databases lack information about human disease for researchers
studying disease-related proteins. We have developed MspecLINE, a tool that combines knowledge about human
disease in MEDLINE with empirical data about the detectable human proteome in PeptideAtlas. MspecLINE
associates diseases with proteins by calculating the semantic distance between annotated terms from a
controlled biomedical vocabulary. Our semantic distance measure is based on the co-occurrence of disease and
protein terms in the MEDLINE bibliographic database.
The MspecLINE web application allows researchers to explore relationships between human diseases
and parts of the proteome that are detectable using a mass spectrometer. Given a disease, the tool will display
proteins and peptides from PeptideAtlas that may be associated with the disease. It will also display relevant
literature from MEDLINE. Furthermore, MspecLINE allows researchers to select proteotypic peptides for speci?c
protein targets in a mass spectrometry assay.
Although MspecLINE applies an information retrieval technique to the MEDLINE database, it is
distinct from previous MEDLINE query tools in that it combines the knowledge expressed in scienti?c literature
with empirical proteomics data. The tool provides valuable information about candidate protein targets to
researchers studying human disease and is freely available on a public web server.
Long Abstract:Click Here

Accepted Posters

View Posters By Category
Search Posters:
Poster Number Matches
Last Name
Co-Authors Contains
Abstract Contains