Processing of Agilent microRNA array data
1 Department of Cardiovascular Epidemiology and Population Genetics, Centro Nacional de Investigaciones Cardiovasculares Carlos III (CNIC), Madrid, Spain
2 Department of Epidemiology, Johns Hopkins Bloomberg School of Public Health, Baltimore, MD, USA
3 Department of Regenerative Cardiology, Centro Nacional de Investigaciones Cardiovasculares Carlos III (CNIC), Madrid, Spain
4 Genomics Unit, Centro Nacional de Investigaciones Cardiovasculares (CNIC), Madrid, Spain
5 Department of Biostatistics, Johns Hopkins Bloomberg School of Public Health, Baltimore, MD, USA
BMC Research Notes 2010, 3:18 doi:10.1186/1756-0500-3-18Published: 22 January 2010
The Agilent microRNA microarray platform interrogates each microRNA with several copies of distinct oligonucleotide probes and integrates the results into a total gene signal (TGS), using a proprietary algorithm that makes use of the background subtracted signal. The TGS can be normalized between arrays, and the Agilent recommendation is either not to normalize or to normalize to the 75th percentile signal intensity. The robust multiarray average algorithm (RMA) is an alternative method, originally developed to obtain a summary measure of mRNA Affymetrix gene expression arrays by using a linear model that takes into account the probe affinity effect. The RMA method has been shown to improve the accuracy and precision of expression measurements relative to other competing methods. There is also evidence that it might be preferable to use non-corrected signals for the processing of microRNA data, rather than background-corrected signals. In this study we assess the use of the RMA method to obtain a summarized microRNA signal for the Agilent arrays.
We have adapted the RMA method to obtain a processed signal for the Agilent arrays and have compared the RMA summarized signal to the TGS generated with the image analysis software provided by the vendor. We also compared the use of the RMA algorithm with uncorrected and background-corrected signals, and compared quantile normalization with the normalization method recommended by the vendor. The pre-processing methods were compared in terms of their ability to reduce the variability (increase precision) of the signals between biological replicates. Application of the RMA method to non-background corrected signals produced more precise signals than either the RMA-background-corrected signal or the quantile-normalized Agilent TGS. The Agilent TGS normalized to the 75% percentile showed more variation than the other measures.
Used without background correction, a summarized signal that takes into account the probe effect might provide a more precise estimate of microRNA expression. The variability of quantile normalization was lower compared with the normalization method recommended by the vendor.