Adjustment method for microarray data generated using two-cycle RNA labeling protocol

BMC Genomics

Research Article

Fugui Wang¹  Dong Ji²  Minping Qian²  Minghua Deng³  Rui Chen⁴  Shunong Bai⁴ 

[1] Center for Quantitative Biology, Peking University, 100871, Beijing, China;Center for Quantitative Biology, Peking University, 100871, Beijing, China;LMAM,School of Mathematical Sciences, Peking University, 100871, Beijing, China;Center for Quantitative Biology, Peking University, 100871, Beijing, China;LMAM,School of Mathematical Sciences, Peking University, 100871, Beijing, China;Center for Statistical Sciences, Peking University, 100871, Beijing, China;School of Life Science, Peking University, 100871, Beijing, China;

关键词: Microarray;    Gene Expression;    IVT;    Two cycle amplification;    RNA Degradation;    Bias correction;    Clustering;   

DOI  :  10.1186/1471-2164-14-31

 received in 2012-05-18, accepted in 2012-12-26,  发布年份 2013

来源: Springer

【 摘 要 】

BackgroundMicroarray technology is widely utilized for monitoring the expression changes of thousands of genes simultaneously. However, the requirement of relatively large amount of RNA for labeling and hybridization makes it difficult to perform microarray experiments with limited biological materials, thus leads to the development of many methods for preparing and amplifying mRNA. It is addressed that amplification methods usually bring bias, which may strongly hamper the following interpretation of the results. A big challenge is how to correct for the bias before further analysis.ResultsIn this article, we observed the bias in rice gene expression microarray data generated with the Affymetrix one-cycle, two-cycle RNA labeling protocols, followed by validation with Real Time PCR. Based on these data, we proposed a statistical framework to model the processes of mRNA two-cycle linear amplification, and established a linear model for probe level correction. Maximum Likelihood Estimation (MLE) was applied to perform robust estimation of the Retaining Rate for each probe. After bias correction, some known pre-processing methods, such as PDNN, could be combined to finish preprocessing. Then, we evaluated our model and the results suggest that our model can effectively increase the quality of the microarray raw data: (i) Decrease the Coefficient of Variation for PM intensities of probe sets; (ii) Distinguish the microarray samples of five stages for rice stamen development more clearly; (iii) Improve the correlation coefficients among stamen microarray samples. We also discussed the necessity of model adjustment by comparing with another simple adjustment method.ConclusionWe conclude that the adjustment model is necessary and could effectively increase the quality of estimation for gene expression from the microarray raw data.

【 授权许可】

Unknown   
© Wang et al.; licensee BioMed Central Ltd. 2013. This article is published under license to BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.