Nagarajan Radhakrishnan, Upreti Meenakshi
Division of Biomedical Informatics, University of Arkansas for Medical Sciences, Little Rock, Arkansas, USA.
Methods Enzymol. 2011;487:133-46. doi: 10.1016/B978-0-12-381270-4.00005-6.
Inferring functional relationships and network structure from the observed gene expression profiles can provide a novel insight into the working of the genes as a system or network as opposed to independent entities. Such networks may also represent possible causal relationships between a given set of genes, hence can prove to be a convenient abstraction of the underlying signaling mechanism. The discovery of functional relationships from the observed gene expression profiles does not rely on prior literature, hence useful in identifying undocumented relationships between a given set of genes. Several techniques have been proposed in the literature. The present study investigates the choice Granger causality (GC) and its extensions in modeling the network structure between a given pair of genes from their expression profiles. The impact of noise variance on GC relationships is investigated. VAR parameter estimation is proposed to obtain a finer insight into the functional relationships inferred using GC tests. The results are presented on synthetic networks generated from known vector-autoregressive (VAR) models and those from cell-cycle gene expression profiles that can be modeled as a first-order bivariate VAR.
从观察到的基因表达谱推断功能关系和网络结构，可以为基因作为一个系统或网络（而非独立实体）的运作方式提供全新的见解。这样的网络也可能代表给定基因集之间可能的因果关系，因此可以证明是对潜在信号传导机制的一种便捷抽象。从观察到的基因表达谱中发现功能关系并不依赖于先前的文献，因此有助于识别给定基因集之间未被记录的关系。文献中已经提出了几种技术。本研究调查了格兰杰因果关系（GC）及其扩展在根据给定基因对的表达谱对网络结构进行建模中的应用。研究了噪声方差对GC关系的影响。提出了向量自回归（VAR）参数估计，以便更深入地了解使用GC检验推断出的功能关系。结果展示在由已知向量自回归（VAR）模型生成的合成网络以及可建模为一阶二元VAR的细胞周期基因表达谱的网络上。
