Microarray image compression: SLOCO and the effect of information loss
Signal Processing - Special issue: Genomic signal processing
Gridding and Compression of Microarray Images
CSB '04 Proceedings of the 2004 IEEE Computational Systems Bioinformatics Conference
| Hi-index | 0.00 |
The ubiquity of microarray expression data in state-of-the-art biology has been well established. The widespread adoption of this technology coupled with the significant volume of image-based experimental data generated per experiment (averaging 40 MB), have led to significant challenges in storage and query-retrieval of primary data from microarray experiments. Research in the yet nascent area of microarray data-compression seeks to address this problem. In this paper, we propose a conceptually novel approach that achieves significantly better lossless compression. Unlike lossy compression, our algorithm is guaranteed against loss of information that may have potential biological relevance. The proposed method supports key operations such as automated grid and spot finding, histogram-based automatic thresholding for spot segmentation, and subsequent foreground and background separation. Based on the proposed approach, we have developed a standardized format for storing microarray data that encapsulates all the relevant information, including both the Cy3 and Cy5 expression images significantly compressed. We have also developed a software application called MACE (Microarray Compression and Extraction application) to compress-decompress microarray data and generate the aforementioned format. Compression-decompression results on a wide class of microarray experiments involving different spot layouts validate the effectiveness of our approach and its potential to significantly address the aforementioned challenges in storage and management of microarray data.