Edit: this was in answer to simultaneous use of robust and weighting methods in limma.

Hi Richard,

In principle, they can be used in any combination, but the effectiveness of this awaits careful testing. I would personally be reluctant to use lmFit(method="robust") with the other methods just because I don't trust the variance estimators from the MM regression that much.

lmFit(method="robust") is designed to deal with individual expression values as outliers. arrayWeights() is designed to deal with outlier arrays.  eBayes(robust=TRUE) is designed to deal with outlier (hypervariable) genes. So the first is observation based, the second is array based, and the third is gene based.  Rather than trying all combinations, I would be guided by the scientific context and what type of aberration seems of high risk. Outlier arrays typically arise when RNA samples vary markedly in quality, and this is common in human clinical studies when RNA is hard to get.  Outlier genes typically arise when a minority of genes are affected by a hidden covariate or batch effect.

lmFit(method="robust") has been in limma since the earliest days, but it hasn't been used so much in practice. This may be because microarrays have a limited dynamic range and so don't tend to show dramatic single-observation outliers. (RNA-seq may prove to be different.) Or it might be because the least squares approach on the log-scale is pretty robust anyway.

Most people might be familiar with robust methods as a way to add protection against outliers, but array or gene outliers tend to produce conservative results in the limma pipeline anyway. The major purpose of arrayWeights and eBayes(robust=TRUE) is to recover statistical power in the presence of poor data, without having to make ad hoc judgements about which poorer quality arrays or probes to remove from the analysis.

Best wishes
Gordon

> Date: Wed, 11 Dec 2013 13:14:24 -0500
> From: Richard Friedman <friedman at c2b2.columbia.edu>
> To: Bioconductor list <bioconductor@r-project.org>
> Subject: [BioC] simultaneous use of robust and weighting methods in limma.
>
> Dear List,
>
> Should arrayweights,  eBayes(robust=TRUE), and lmFit(...,method="robust")
> be used simulatanenously in Limma? If not should any combination be used
> together?
>
> Thanks and best wishes,
> Rich
> Richard A. Friedman, PhD
> Associate Research Scientist,
> Biomedical Informatics Shared Resource
> Herbert Irving Comprehensive Cancer Center (HICCC)
> Lecturer,
> Department of Biomedical Informatics (DBMI)
> Educational Coordinator,
> Center for Computational Biology and Bioinformatics (C2B2)
> National Center for Multiscale Analysis of Genomic Networks (MAGNet)
> Columbia Department of Systems Biology
> Room 824
> Irving Cancer Research Center
> Columbia University
> 1130 St. Nicholas Ave
> New York, NY 10032
> (212)851-4765 (voice)
> friedman at c2b2.columbia.edu
> http://friedman.c2b2.columbia.edu/
>
> In memoriam, Frederik Pohl