A Two-Stage Procedure for the Removal of Batch Effects in Microarray Studies

The presence of different batches is routinely observed in microarray studies and is well known that non-biological variability potentially confounding biological differences is commonly related to such batches. The removal of these undesired effects for a non-biased inference is often accomplished either with normalization methods that do not take into account all the available information, or with models that rely on strong parametric assumptions. We have developed a new method for the batch effects removal, named ber, which is based on a two-stage procedure for the estimation of location and scale parameters. Batch effects and biological differences are estimated using a regression approach and bagging, therefore mild distributional assumptions are required. We have compared ber with other commonly employed methods and we have shown that ber can bring to a higher power in detecting differentially expressed genes. The application of ber to a real microarray study led to interpretable biological results. The method is implemented in the R package ber, available through CRAN repository.

[1]  Leo Breiman,et al.  Bagging Predictors , 1996, Machine Learning.

[2]  Andrew B. Nobel,et al.  Merging two gene-expression studies via cross-platform normalization , 2008, Bioinform..

[3]  B. Efron.  Large-Scale Simultaneous Hypothesis Testing , 2004 .

[4]  Rafael A Irizarry,et al.  Frozen robust multiarray analysis (fRMA). , 2010, Biostatistics.

[5]  H. Glejser.  A New Test for Heteroskedasticity , 1969 .

[6]  R. Irizarry,et al.  A gene expression bar code for microarray data , 2007, Nature Methods.

[7]  Korbinian Strimmer,et al.  A unified approach to false discovery rate estimation , 2008, BMC Bioinformatics.

[8]  Cheng Li,et al.  Adjusting batch effects in microarray expression data using empirical Bayes methods. , 2007, Biostatistics.

[9]  Chunyu Liu,et al.  Removing Batch Effects in Analysis of Expression Microarray Data: An Evaluation of Six Batch Adjustment Methods , 2011, PloS one.

[10]  Rafael A Irizarry,et al.  Exploration, normalization, and summaries of high density oligonucleotide array probe level data. , 2003, Biostatistics.

[11]  H. Joe.  Generating random correlation matrices based on partial correlations , 2006 .

[12]  A. Zinober.  Matrices: Methods and Applications , 1992 .

[13]  Martin Vingron,et al.  Variance stabilization applied to microarray data calibration and to the quantification of differential expression , 2002, ISMB.

[14]  L. Bullinger,et al.  Gene expression profiling in AML with normal karyotype can predict mutations for molecular markers and allows novel insights into perturbed biological pathways , 2010, Leukemia.

[15]  Anne-Laure Boulesteix,et al.  CMA – a comprehensive Bioconductor package for supervised classification with high dimensional data , 2008, BMC Bioinformatics.

[16]  Matthew N. McCall,et al.  The Gene Expression Barcode: leveraging public data repositories to begin cataloging the human and murine transcriptomes , 2010, Nucleic Acids Res..

[17]  M. J. van de Vijver,et al.  Gene expression profiling in breast cancer: understanding the molecular basis of histologic grade to improve prognosis. , 2006, Journal of the National Cancer Institute.

[18]  John D. Storey,et al.  Supervised normalization of microarrays , 2010, Bioinform..

[19]  Korbinian Strimmer,et al.  An empirical Bayes approach to inferring large-scale gene association networks , 2005, Bioinform..

[20]  K. Strimmer,et al.  Statistical Applications in Genetics and Molecular Biology A Shrinkage Approach to Large-Scale Covariance Matrix Estimation and Implications for Functional Genomics , 2011 .

[21]  Joel S. Parker,et al.  Adjustment of systematic microarray data biases , 2004, Bioinform..

[22]  Sean R. Davis,et al.  GEOquery: a bridge between the Gene Expression Omnibus (GEO) and BioConductor , 2007, Bioinform..