Fan Xiaohui, Fang Hong, Hong Huixiao, Perkins Roger, Shi Leming, Tong Weida
National Center for Toxicological Research, US Food and Drug Administration, Jefferson, AR 72079, USA.
Anal Biochem. 2009 Feb 15;385(2):203-7. doi: 10.1016/j.ab.2008.11.019. Epub 2008 Dec 6.
Quality control of a microarray experiment has become an important issue for both research and regulation. External RNA controls (ERCs), which can be either added to the total RNA level (tERCs) or introduced right before hybridization (cERCs), are designed and recommended by commercial microarray platforms for assessment of performance of a microarray experiment. However, the utility of ERCs has not been fully realized mainly due to the lack of sufficient data resources. The US Food and Drug Administration (FDA)-led community-wide Microarray Quality Control (MAQC) study generates a large amount of microarray data with implementation of ERCs across several commercial microarray platforms. The utility of ERCs in quality control by assessing the ERCs' concentration-response behavior was investigated in the MAQC study. In this work, an ERC-based correlation analysis was conducted to assess the quality of a microarray experiment. We found that the pairwise correlations of tERCs are sample independent, indicating that the array data obtained from different biological samples can be treated as technical replicates in analysis of tERCs. Consequently, the commonly used quality control method of applying correlation analysis on technical replicates can be adopted for assessing array performance based on different biological samples using tERCs. The proposed approach is sensitive to identifying outlying assays and is not dependent on the choice of normalization method.
微阵列实验的质量控制已成为研究和监管中的一个重要问题。外部RNA对照（ERC），既可以添加到总RNA水平（tERC），也可以在杂交前直接引入（cERC），是商业微阵列平台设计并推荐用于评估微阵列实验性能的。然而，由于缺乏足够的数据资源，ERC的效用尚未得到充分实现。由美国食品药品监督管理局（FDA）牵头的全社区微阵列质量控制（MAQC）研究通过在多个商业微阵列平台上实施ERC生成了大量微阵列数据。MAQC研究中对通过评估ERC的浓度-反应行为来进行质量控制时ERC的效用进行了研究。在这项工作中，进行了基于ERC的相关性分析以评估微阵列实验的质量。我们发现tERC的成对相关性与样本无关，这表明从不同生物样本获得的阵列数据在tERC分析中可被视为技术重复。因此，基于tERC对不同生物样本进行阵列性能评估时，可以采用在技术重复上应用相关性分析这种常用的质量控制方法。所提出的方法对于识别异常检测很敏感，并且不依赖于归一化方法的选择。
