MICROARRAY EXPERIMENTS : APPLICATION TO SPORULATION TIME SERIES
MICROARRAY EXPERIMENTS : APPLICATION TO SPORULATION TIME SERIES
复制标题
DOI：
--
复制
发表时间：
1999
期刊：
影响因子：
--
通讯作者：
S. Raychaudhuri;Joshua M. Stuart;R. Altman
中科院分区：
文献类型：
--
作者：
S. Raychaudhuri;Joshua M. Stuart;R. Altman
微阵列实验产生的大量数据可能很难处理。给定的一系列微阵列实验在多种条件下观察到数千个基因的差异表达。这些大型数据集可以用主成分分析（PCA）进行总结，这是一种统计技术，可以识别多维数据集中的关键变量（或变量组合）。主成分分析确定数据中最能解释观测差异的关键变量。在这里，我们展示了将PCA应用于表达数据的实用性，其中实验条件是变量，基因表达测量是观察结果。因此，每个组件定义了一个线性组合的实验条件，可以用来区分基因吝啬。对这些组成部分的检查还可以深入了解实验中实际测量的潜在因素。我们将PCA应用于公开发布的酵母孢子形成数据集（Chu et al. 1998）。在这项工作中，随着时间的推移，对基因表达进行了7种不同的测量。时间点的主成分分析表明，实验中观察到的大部分变异性可以总结为2个分量，即2个变量捕获了大部分信息。这些潜在因素似乎代表（1）总体诱导水平和（2）诱导水平随时间的变化。我们的结果的可视化可供使用（http：//www.smi.stanford.edu/projects/helix/PCArray）。
The enormous amount of data produced by microarray experiments can be unwieldy. A given series of microarray experiments produces observations of differential expression for thousands of genes across multiple conditions. These large data sets can be summarized with principal components analysis (PCA), a statistical technique that allows the key variables (or combinations of variables) in a multidimensional data set to be identified. Principal components analysis determines those key variables in the data that best explain the differences in the observations. Here we show the utility of applying PCA to expression data, where the experimental conditions are the variables, and the gene expression measurements are the observations. Thus, each component defines a linear combination of the experimental conditions that can be used to distinguish genes parsimoniously. Examination of the components also provides insight into what underlying factors are actually being measured in the experiment. We applied PCA to the publicly released yeast sporulation data set (Chu et al. 1998). In that work, 7 different measurements of gene expression were made over time. PCA on the time-points suggests that much of the observed variability in the experiment can be summarized in just 2 components—i.e. 2 variables capture most of the information. These underlying factors appear to represent (1) overall induction level and (2) change in induction level over time. A visualization of our results is made available (http://www.smi.stanford.edu/projects/helix/PCArray).