Sidekick: Integrating knowledge and user belief to enable biological discovery

Date
2010
Authors
Doderer, Mark S.
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract

Scientists striving to unlock mysteries within complex biological systems face myriad barriers in effectively integrating available information to enhance their understanding. Useful information is dispersed across a variety of sources, and sources of the same information often do not use the same format or nomenclature. Scientists need tools that bridge nomenclature differences and allow them to integrate, organize, and evaluate the quality of information without extensive computation.

Sidekick, a genomic data-driven analysis and decision-making application, is a web-based collection of tools to query for information, combine, manage and evaluate the information and discover connections among the information. Unlike a workflow with a series of prescribed steps, the application of tools provides a flexible solution to the problem of knowledge discovery. Sidekick enables scientists without training in computation and data management to pursue research questions like "Are there other mechanisms for disease X" or "Does the set of genes associated with disease X also influence other diseases" without prescribing the steps to answer the questions. Sidekick enables the process of combining heterogeneous data, finding and maintaining the most up-to-date data, evaluating data sources, quantifying confidence in results based on evidence, and managing the multi-step research tasks needed to answer these questions. Possible analysis steps include gene list discovery, gene-pair list discovery, and enrichment for both types of lists.

Similar tools exist for some of these tasks, however Sidekick's Belief Manager provides a new visualization approach for quantifying confidence in results based on evidence and user belief. Also, Sidekick's ability to characterize pairs of genes offers a new way to understand gene relationships that traditional single gene lists do not, particularly in areas such as interaction discovery. We demonstrate Sidekick's use of these new tools and overall effectiveness and flexibility by showing how to accomplish a complex published analysis involving protein-protein interactions in a fraction of the original time; while encapsulating the computational work and hiding it from the Sidekick user.

Description
This item is available only to currently enrolled UTSA students, faculty or staff.
Keywords
belief, biological, discovery, knowledge
Citation
Department
Computer Science