Reverter Antonio, Ingham Aaron, Lehnert Sigrid A, Tan Siok-Hwee, Wang Yonghong, Ratnakumar Abhirami, Dalrymple Brian P
CSIRO Livestock Industries, Queensland Bioscience Precinct 306 Carmody Road, Brisbane, Queensland 4067, Australia.
Bioinformatics. 2006 Oct 1;22(19):2396-404. doi: 10.1093/bioinformatics/btl392. Epub 2006 Jul 24.
Biological differences between classes are reflected in transcriptional changes which in turn affect the levels by which essential genes are individually expressed and collectively connected. The purpose of this communication is to introduce an analytical procedure to simultaneously identify genes that are differentially expressed (DE) as well as differentially connected (DC) in two or more classes of interest.
Our procedure is based on a two-step approach: First, mixed-model equations are applied to obtain the normalized expression levels of each gene in each class treatment. These normalized expressions form the basis to compute a measure of (possible) DE as well as the correlation structure existing among genes. Second, a two-component mixture of bi-variate distributions is fitted to identify the component that encapsulates those genes that are DE and/or DC. We demonstrate our approach using three distinct datasets including a human systemic inflammation oligonucleotide data; a spotted cDNA data dealing with bovine in vitro adipogenesis and SAGE database on cancerous and normal tissue samples.
类别之间的生物学差异反映在转录变化中，而转录变化又会影响必需基因的个体表达水平和整体关联水平。本通讯的目的是介绍一种分析程序，以同时识别在两类或更多感兴趣的类别中差异表达（DE）以及差异关联（DC）的基因。
我们的程序基于两步法：首先，应用混合模型方程来获得每个基因在每种类别处理中的标准化表达水平。这些标准化表达构成了计算（可能的）DE度量以及基因之间存在的相关结构的基础。其次，拟合双变量分布的双组分混合物，以识别包含那些DE和/或DC基因的组分。我们使用三个不同的数据集展示了我们的方法，包括人类全身炎症寡核苷酸数据；处理牛体外脂肪生成的斑点cDNA数据以及癌组织和正常组织样本的SAGE数据库。
