Time ordering of gene coexpression
Time ordering of gene coexpression
复制标题
DOI：
10.1093/biostatistics/kxj026
复制
发表时间：
2006-10-01
期刊：
影响因子：
2.1
通讯作者：
Mueller, Hans-Georg
中科院分区：
文献类型：
--
作者：
Leng, Xiaoyan;Mueller, Hans-Georg
时间微阵列基因表达谱可以通过同一遗传途径中基因共表达的时间动态来表征基因功能。在这篇文章中，我们通过最小二乘法定义和估计每个基因的全局时移特征，从两条曲线的比对中推断出来。单个基因的这些时移特征反映了从观察到的时间基因表达谱得出的时间顺序。一旦获得了每个基因的这些时移特征，它们就可以进入进一步的分析，如聚类。我们使用果蝇胚胎发育和酵母细胞周期基因表达谱以及模拟来说明所提出的方法。通过对时间序列的成功恢复，验证了该算法的可行性。果蝇母体基因和合子基因的估计时间漂移提供了这两个类别之间的出色区分，并通过基因表达的时间顺序确认了已知的遗传途径。对酵母细胞周期数据的应用建立了与细胞周期相一致的基因的自然时间顺序。该方法不需要基因表达谱的周期性。文中还给出了渐近证明。
Temporal microarray gene expression profiles allow characterization of gene function through time dynamics of gene coexpression within the same genetic pathway. In this paper, we define and estimate a global time shift characteristic for each gene via least squares, inferred from pairwise curve alignments. These time shift characteristics of individual genes reflect a time ordering that is derived from ob- served temporal gene expression profiles. Once these time shift characteristics are obtained for each gene, they can be entered into further analyses, such as clustering. We illustrate the proposed methodology using Drosophila embryonic development and yeast cell-cycle gene expression profiles, as well as simulations. Feasibility is demonstrated through the successful recovery of time ordering. Estimated time shifts for Drosophila maternal and zygotic genes provide excellent discrimination between these two categories and confirm known genetic pathways through the time order of gene expression. The application to yeast cell-cycle data establishes a natural time order of genes that is in line with cell-cycle phases. The method does not require periodicity of gene expression profiles. Asymptotic justifications are also provided.