Searching the principal genes for neural differentiation of mouse ES cells by factorizing eigengenes of clusters |
| |
Authors: | Kim Hye Young Kim Min Jung Han Jin Il Kim Bo Kyung Lee Young Seek Lee Yong Sung Kim Jin Hyuk |
| |
Institution: | Department of Physiology, College of Medicine, Hanyang University, Seoul 133-791, Republic of Korea. hykim121@hanyang.ac.kr |
| |
Abstract: | A time-series microarray experiment is useful to study the changes in the expression of a large number of genes over time. Many methods for clustering genes using gene expression profiles have been suggested, but it is not easy to interpret the biological significance of the results or utilize these methods for understanding the dynamics of gene regulatory systems. In this study, we introduce an algorithm for readjusting the boundaries of clusters by adopting the advantages of both k-means and singular value decomposition (SVD). In addition, we suggest a methodology for searching the principal genes that can be the most crucial genes in regulation of clusters. We found 34 principal genes from 171 clusters having strong concentratedness in their expression patterns and distinct ranges of oscillatory phases, by using a time-series microarray dataset of mouse embryonic stem (ES) cells after induction of dopaminergic neural differentiation. The biological significance of the principal genes examined in the literature supports the feasibility of our algorithms in that the hierarchy of clusters may lead the manifestation of the phenotypes, e.g., the development of the nervous system. |
| |
Keywords: | Principal gene Eigengene Singular value decomposition Moore–Penrose pseudoinverse Depth first search Differentiation Mouse ES cells |
本文献已被 ScienceDirect PubMed 等数据库收录! |
|