Hi,

In the Nat Prot paper about Ballgown, it's said:

Note that Ballgown’s statistical test is a standard linear model-based comparison. For small sample sizes (n < 4 per group), it is often better to perform regularization. This can be done using the limma33 package in Bioconductor.

So, how this can be done? First,by regularization, does it mean a normalization or something more complicated (i'm not deep in statistics)? And second, how can the 'regularized' new data be included in the ballgown object?

When I try this:

bg = ballgown(dataDir = "ballgown", samplePattern = "mRNA", pData=pheno_data, meas="all")
texpr(bg) <- normalizeBetweenArrays(texpr(bg))

I got an error because I cannot modify the transcript matrix (and also as I said, I don't know if normalizeBetweenArrays is the proper regularization, I guess no)

Thanks!

The idea behind the sentence in the paper about performing regularization for small sample sizes is basically a suggestion to use the expression data from ballgown as the inputs to limma. limma requires you to generate an expression matrix, and if you would like to use their methods, you would use ballgown to get that expression matrix (with texpr, or something similar), but then from there on, you'd use the limma package. (there is no "texpr <- " method in ballgown; texpr is the core of the ballgown object so you cannot overwrite it. If you would like regularized data, you should store it as a separate object which would then be passed to the appropriate limma functions).