[BioC] Treatment of Duplicate spots

[Matthew Ritchie]

Hi Peter,

> I have 3 questions:
>
> How are duplicate spots treated when performing the 
> normalizationWithinArrays? Are they treated separately as far as the 
> regression is concerned?

> How are duplicate spots treated when performing the 
> normalizationBetweenArrays? Are they somehow treated together in the 
> scale normalization or independently? For the case of using quantile 
> normalization, is the number of quantiles the total number of spots on 
> the chip, or it is for the case of duplicate spotting, the number of 
> quantiles are n-spots/2, where each pair are adjusted together in some 
> way?

The duplicate spots are kept separate in the normalization functions 
normalizeWithinArrays() and normalizeBetweenArrays().
Normalization is intended to remove systematic biases from the data.  
Some effects might vary locally on the array, affecting one of the 
duplicates but not the other, so it would seem reasonable to keep the 
duplicates separate so that the normalization has a chance to correct 
such a bias.

The information from the duplicate spots can be summarised using lmFit() 
with the appropriate arguments.  The approach taken in limma is to 
assume that the duplicate spots are correlated by being on the same 
array, a fixed distance apart (the function duplicateCorrelation() is 
used to estimate this correlation).  An alternative approach would be to 
average the duplicate log-ratios prior to fitting the linear model.

> For the case of duplicate spotting, what is the significance of 
> merging the raw channels seperately prior to creating MA values with 
> the loess normalization, then between chip scaling.

I'm not sure what you mean here.  There are usually two channels per 
array for two-colour microarrays.  Do you mean create 4 channels per 
array, one for each duplicate set in each channel?  I'm not sure that 
this would be helpful.

> How many spots in a chip would be required to run quantile 
> normalization vs scale normalization when using normalizeBetweenArrays?

The lower limit for quantile normalization is 2 spots, and for scale 
normalization it's 1 spot.  Normalization is probably not such a big 
deal with so few spots though ;)

> Thanks for any insight into this. I am not a statistician, so I am 
> unfamiliar with the ramifications of duplicate treatment in regression 
> and in normalization.
>
> Peter W.

Best wishes,

Matt Ritchie