XMAS: An Experiential Approach for Visualization, Analysis, and Exploration of Time Series Microarray Data

Abstract

Time series microarray analysis provides an invaluable insight into the genetic progression of biological processes, such as pregnancy and disease. Many algorithms and systems exist to meet the challenge of extracting knowledge from the resultant data sets, but traditional methods limit user interaction, and depend heavily on statistical, black box techniques. In this paper we present a new design philosophy based on increased human computer synergy to overcome these limitations, and facilitate an improved analysis experience. We present an implementation of this philosophy, XMAS (eXperiential Microarray Analysis System) which supports a new kind of “sit forward” analysis through visual interaction and interoperable operators. Domain knowledge, (such as pathway information) is integrated directly into the system to aid users in their analysis. In contrast to the “sit back”, algorithmic approach of traditional systems, XMAS emphasizes interaction and the power, and knowledge transfer potential of facilitating an analysis in which the user directly experiences the data. Evaluation demonstrates the significance and necessity of such a philosophy and approach, proving the efficacy of XMAS not only as tool for validation and sense making, but also as an unparalleled source of serendipitous results. Finally, one can download XMAS at http://cose-stor.sfsu.edu/~huiyang/ xmas_website/xmas.html

1 Introduction

Microarray-based experimentation is a technique, which measures the expression levels for hundreds and thousands of genes within a tissue or cell simultaneously. It therefore provides a data rich environment to obtain a systemic understanding of various biochemical processes and their interactions. Data from microarray experiments have been used, among others, to infer probable functions of known or newly discovered genes based on similarities in expression patterns with genes of known functionality, reveal new expression patterns of genes across gene families, and even uncover entirely new categories of genes [1], [2]. In more applied settings, microarray data has provided biologist with ways of identifying genes that are implicated in various diseases through the comparison of expression patterns in diseased and healthy tissues.

This research was partially funded by a grant from the National Science Foundation IIS-0644418 (CAREER) – Singh; SFSU CCLS Mini Grants – Yang.