XMAS: An Experiential, Time Series Microarray Data Analysis System
Benjamin DalzielOral Defence Date:
Friday, June 19, 2009 - 18:00Location:
Professors Singh, Yang and Petkovic
Time series microarray studies provide an opportunity to obtain invaluable insight into a wide range of biological processes such as pregnancy, drug treatment, and disease. Many algorithms, techniques and software systems have been developed to extract knowledge from the data obtained from such studies, but traditional methods provide limited user interaction and depend heavily on statistical, black box techniques. To overcome such limitations, we have designed and implemented XMAS (eXperiential Microarray Analysis System), a software system for time-series microarray data visualization, analysis, knowledge discovery, and hypothesis formulation. The design paradigm underlying XMAS facilitates a harmonious synergy between human and computer, where the human skills of domain knowledge application, contextual reasoning, and purpose-driven exploration are seamlessly combined with computable operations that support large-scale data analysis, multifaceted data visualization, and multi-source data integration. This synergy allows a novel form of “sit forward” analysis where visual interaction and interoperable operators, which can be invoked on-demand, allow the user to focus on data exploration, hypothesis formulation, and discovery of salient patterns of interest. Simultaneously, the user is freed from the tedium of tasks that are computational, repetitive, or focus on data integrative aspects. This visualization-interaction-exploration strategy of XMAS differs from both the traditional algorithm-centric approaches as well as pure visualization systems where interactive access to powerful algorithmic components is limited. The main contributions of this work include: (1) increased user involvement, comprehension and understanding through development of a new design philosophy for microarray data analysis; (2) improved biological results from analysis; and (3) a concrete web-based extensible implementation of this design philosophy. Detailed case studies and user studies are presented to prove the efficacy of the solution. XMAS is available both as a hosted web application and as a downloadable, locally deployable version at: http://tintin.sfsu.edu:8080/xmas. XMAS is an open source project.