Reverse Engineering Gene Regulatory Networks with Various Machine Learning Methods
Reverse Engineering Gene Regulatory Networks with Various Machine Learning Methods
复制标题
DOI：
10.1002/9783527622818.ch5
复制
发表时间：
2008-01-01
期刊：
影响因子：
--
通讯作者：
Werhli, Adriano V.
中科院分区：
文献类型：
--
作者：
Grzegorczyk, Marco;Husmeier, Dirk;Werhli, Adriano V.
系统生物学中的一个重要问题是从后基因组数据推断生化途径和调控网络。文献中提出了各种逆向工程方法，了解它们的优缺点是很重要的。在本章中，我们比较了三种不同的建模和推理范例重建基因调控网络的准确性：(1)相关性网络(RN)：独立于剩余网络的成对关联分数；(2)图形高斯模型(GGMS)：基于约束推理的无向图形模型；(3)贝叶斯网络(BNS)：基于分数推理的有向图形模型。在简要介绍了这三种方法的方法学概念之后，我们提出了一个使用细胞学实验的实验室数据和模拟数据的比较评估。我们的评估是基于一个已发表的金标准细胞调控网络，该网络描述了11种磷酸化蛋白质和磷脂在人类免疫系统细胞中的相互作用。在我们的研究中，特别感兴趣的是被动观测和主动干预之间的比较，以及从后者获得的网络重建精度的改善。
An important problem in systems biology is the inference of biochemical pathways and regulatory networks from postgenomic data. Various reverse engineering methods have been proposed in the literature, and it is important to understand their relative merits and shortcomings. In the present chapter, we compare the accuracy of reconstructing gene regulatory networks with three di®erent modelling and inference paradigms: (1) Relevance networks (RNs): pairwise association scores independent of the remaining network; (2) graphical Gaussian models (GGMs): undirected graphical models with constraint-based inference, and (3) Bayesian networks (BNs): directed graphical models with score-based inference. After providing a concise and self-contained introduction to the methodological concepts of these three approaches, we present a comparative evaluation using both laboratory data from cytometry experiments as well as simulated data. Our evaluation is based on a published gold-standard cellular regulatory network describing the interaction of eleven phosphorylated proteins and phospholipids in human immune system cells. Of particular interest in our study is a comparison between passive observations and active interventions, and a quanti¯cation of the improvement in network reconstruction accuracy obtained from the latter.