A systematic approach of modulization in hES cell differentiation

초록

Due to the natural complexities of organisms, it is difficult to elucidate the principal mechanisms that lead to a specific phenomenon. A biological phenomenon is observed as a mixture state of the various orthogonal biological events. Therefore a systematic view of the biological processes is required to investigate the mechanisms that operate the orthogonal biological events to generate the complicated biological processes. In the genetic networks, a genetic module is a functional unit composed of genes that perform a specific activity together upon the regulatory information. Therefore, for a given module, the information between genes satisfies the max-flow, and a total amount of the information transmitting to the genetic module is extremely limited. However, since the existing clustering methods have little consideration for systematic properties of organism, there are limits to approach the genetic modules. In this study, to reverse-engineering the genetic modules, we proposed a novel method for clustering genes, based on the decomposition of the gene expression matrix, and this algorithm was applied to a time-series microarray experiment from human embryonic stem (hES) cells differentiated to dopaminergic (DA) neurons. When we used this algorithm to the differentiated hES cell, genes that have similar function were clustered in the same cluster, and several key genes were identified from the clusters. Therefore, as a first step of understanding differentiation, our method can not only extract significant meanings, but also provides hints to recognize the complex properties in differentiation processes.