Computatonal Analysis of Gene Expression Pattern Images

点击翻译按钮获取中文摘要

DESCRIPTION (provided by applicant): Metazoans are the most complex life forms in which the complexity of body form is achieved by spatial and temporal patterns of gene expression and its regulation in the developing embryo. The Drosophila melanogaster has been a model system to study how this complexity arises by examining the gene expression patterns depicting the mRNA or protein localization of key genes in the developing embryo. Investigators exploring a specific set of genes or gene families are often interested in finding other genes with overlapping patterns of gene expression in order to elucidate developmental pathways. However, neither a biological database of these gene expression pattern images nor a computational framework exists (1) to find similar gene expression patterns, and (2) to access knowledge related to the known genetic interactions. Therefore, we propose to build a fruit-fly gene expression system (FlyExpress) for retrieval and visualization of spatial and temporal patterns of gene expression in Drosophila embryos and imaginal disks. The proposed bioinformatics framework will facilitate query for genes with similar expression patterns in the temporal, spatial, and organ specific contexts. In addition to text search based on biological attributes, we will develop a Basic Expression Search Tool visual-content query system to assist in the discovery of overlapping gene expression patterns (Fx-BEST). The Fx-BEST functionality will be analogous to that of BLAST search in molecular sequence analysis. We also plan to develop computational tools to generate organ and position specific gene expression pattern classes (Fx-Classify), where expression patterns showing significant overlap will be grouped in the same class. The resulting classification will allow expert investigators to use biological, developmental, and genetic attributes of these expression patterns to predict potentially new interactions among genes or members of developmental pathways. In addition, we plan to devise computational approaches to compare (in a manner similar to that done by developmental biologists) the expression pattern of a given gene in wild type and mutation backgrounds in order to infer the underlying genetic interactions (Fx-Interaction). This technique will be used to automatically generate all gene interactions constructed using the data curated in the Fx- Database. These interactions can be used to establish further hypotheses for testing in research laboratories. FlyExpress will be made freely accessible on the web (www.flyexpress.net). It will not only address the day-to-day needs of researchers, but will also provide a framework for further discovery using the existing knowledge (approximately 50,000 images and their attributes). Furthermore, the computational algorithms, statistical methods, and the bioinformatics technologies developed in this project will be useful and provide impetus for constructing similar frameworks for organizing gene expression pattern data from other model and non-model organisms. The proposed FlyExpress informatics system will directly facilitate basic and applied research in many areas of molecular biology crucial in human health research, including computational genomics, molecular genetics, developmental biology, genetics, and evolution. It is also likely to become a valuable teaching resource for undergraduate and graduate students at universities worldwide