Development of a gene expression analysis method for large scale sequence data

Abstract

Recent advances in sequencing technologies have led to a remarkable increase in the number of sequences of expressed genes. Comparisons of large-scale sequence data from different samples facilitate identification of such as stage- or tissue-specifically expressed genes. In current approaches for discovering genes from sequence (EST) data, genes are categorized with hierarchical clustering analysis on the basis expression parterres estimated from EST assembling. However, the lack of memory with long calculation time is often caused by large-scale clustering analysis. An innovative method to perform expression analysis with common computational system and short calculation time should be developed to face of the novel sequencing technology. We have developed a new statistical method for large-scale gene expression data by using 'correspondence analysis (CA)'. Sample-specifically expressed genes or genes expressed in some samples are readily detected by the developed method. For results from CA, our three-dimensional plot viewer graphically shows similarities of gene expression parterres. We also present software for expression parterres analysis and interactive plot viewer.