Data sharing, transmission and storage are important problems in gene expression data collection and analysis, due to the large quantity and high cost. We present in this talk a lossless compression scheme for microarray expression data that can achieve a compression ratio of 1:25. Furthermore, the design of this compression package focuses on data exploration and description. Feature components of the compression routine are segmentation and feature extraction, ROI (region of interest) boundary descriptions, background processing, denoising and data quality assessment. Our image compression tool also provides a multi-scale description of the expression data, and researcher can chose to view the images at various levels of detail. We therefore believe that this tool should greatly assist researchers with data exploration. (http://www.stat.berkeley.edu/~binyu)
Copyright. All Rights Reserved.