Signature analysis tools for the life sciences

点击翻译按钮获取中文摘要

Cells have evolved a set of complex, inter-connected processes that perform a wide variety of functions. When a cell is stressed by an environmental challenge or genetic defect, it activates specific processes that attempt to circumvent the problem. Processes are essentially a list of genes or gene products that interact in special ways to accomplish their task. In biological and biomedical research, there is a need to identify if a specific molecular pathway or process is active within a target cell. It is still very hard to do this in the laboratory for even a single process, and it is impossible with current technologies to directly measure the status of all pathways and processes simultaneously. The use of high-throughput technologies such as microarrays and next generation sequencing has provided researchers with a type of experiment that allows all processes within the cell to be identified but it does so in an indirect way. For example, microarrays can be used to measure the expression levels of all genes in the cell simultaneously. If a specific process is active in the cell, it is highly likely that this process will eventually change the expression of some genes. With the use of computers and statistical techniques, these profiles can be searched for evidence of this "signature" of the process. That is, it will identify those genes that are changed when the process is activated. This proposal is aimed at developing bioinformatics/biostatistics methodology that address the many technical difficulties associated with identifying signatures. Our goal is to produce software that helps biologists identify which process are active in their data. In the recent past, systems biology research has established beyond a doubt that cells react to challenges in broad, comprehensive ways, modulating many processes simultaneously in order to survive. Our approach for identifying signatures allows us to form so-called signature networks. These networks provide clues as to how different processes are interconnected and co-modulated. One long-term application may be to assist clinicians in reliably determining those tumors that have a set of processes firing that play a protective role against tumors (thus potentially sparing the patient painful and expensive chemotherapy).