09/15 Dave Newman, Research Scientist Department of Computer Science University of California, Irvine, Gully Burns Research Professor Information Sciences Institute University of Southern California and Bruce Herr II, Visualization Software Developer, Indiana University

Topic Mapping Tools for Biomedical Corpora

Abstract: Biomedical research is supported by several medium-scale corpora that provide a uniform interface to biomedical documents across many subdisciplines. These include (A) Medline (providing access to citations and abstracts for almost all papers published from the 1950s to present), (B) CRISP (providing access to abstracts from NIH-funded proposals); (C) abstracts from specific large-scale conferences (such as the Society for Neuroscience) and (D) full-text collections (such as ScienceDirect from Elsevier). This presentation describes an approach to the analysis of the text within these corpora to provide navigable web-based maps of these document collections that are intuitive and easy to navigate for end-users (administrators, biologists, and doctors). The underlying analysis is based on Latent Dirichlet Allocation (Topic Modeling) which scales well to collections of millions to billions of documents. We use large-scale graph visualization techniques to build a map of the documents within a corpus using a scalable force-directed layout algorithm. This map then forms the basis of a Google Maps user interface that has additional web-support to describe individual documents within the collection. This work (currently supported by NIH) will create tools that allow scientists to evaluate grants from the CRISP database.

10/06 Zhenjun Hu, Bioinformatics Program, Boston University

Multiple-Scale Visualization and Modeling of Biological Networks/Pathways

Abstract: Tools for mining and visualizing cell systems has moved beyond static pictures of networks and links, and now capture functional hierarchies and adaptive networks. Integrative frameworks play a critical role in meeting these challenges. The concept of multiple information scales—for example, the protein molecules that form a complex, the participation of that complex in a pathway, the emergence of phenotype from this pathway, and so on—is central to formulating a global view of network dynamics. Here we present a new graph structure –the metagraph, that is able to integrate the context (temporal or modular activity) and the hierarchical organization of cellular agents (molecules and complexes with associated properties and states) in addition to their interactions, with increased performance and network readability. The features of this new type of graph, as well as its applications and implementations in VisANT will be discussed in detail. VisANT is freely available at http://visant.bu.edu.

See also his paper entitled Towards zoomable multidimensional maps of the cell.

10/27 Faculty, Staff, and Students at IUB

Open House

Abstract: Open your laptops and demo your software. Bring posters to introduce your research questions and results. So far, the following posters and demo's are planned:

• Longitudinal Analysis of Mobility within the American Legal Academy-1922 to 1989: Visualizations, Network Dynamics, Trends, and Emergent Hierarchies by Peter A. Hook
• Preserving e-Science - Perceptions from the Lab Floor by Stacy Kowalczyk
• Stencil: A Declarative, Generative System for Visualizing Streaming Data by Joseph Cottam and Andrew Lumsdaine
• Effects of Social Organization on Information Transfer in a Zebrafish Group by Cuau Vital & Emilia P.Martins,
• TBD by Margaret Clements

See also IV/CNS Open House web site at http://ella.slis.indiana.edu/~katy/gallery/08-openhouse/

11/03 Jon Duke, NLM Medical Informatics Fellow, Regenstrief Institute, Inc., Indianapolis.

Illuminating the Fine Print: Visualizing Medication Side-Effects in Complex Multi-drug Regimens

Abstract: Prescription drug use has increased markedly over the past several decades, with nearly half of Americans over 65 taking at least five medications daily. As these numbers grow, physicians are faced with the increasingly complex task of recognizing and addressing any adverse reactions associated with these treatments. Although information on drug side-effects is readily available, its sheer volume can be daunting: The average drug label contains over 75 potential reactions, and there are many drugs that report well over two-hundred. In this talk, I will discuss the development of an electronic tool which synthesizes adverse reaction data and provides doctors with clinically useful visualizations at the point of care. We will cover many of the challenges in creating such a system, including integration of quantitative and qualitative data, evaluation and iteration of the visualization approach, and actual implementation into physician workflow.