[BioC] Limma: How to analyze 1- and 2- colour Agilent

[Gordon K Smyth]

Dear Edwin,

Your instinct is right I think.  First normalizeWithinArrays on the two 
colour arrays.  Then convert the MAList to a single channel EList (first 
RG.MA then rbind the red and green matrices).  Then rbind the 1- and 2- 
colour experiments.  Then quantile normalization.  Then setup a blocking 
variable that indexes the array that each column of data came from.  In 
this blocking variable, the index of each two colour array will appear 
twice, and each one colour array will appear once.  Then run 
duplicateCorrelation on your data with this blocking variable.  Then run 
lmFit with the blocking variable and the consensus correlation.  This 
approach allows all your data to be analysed, but keeps track of which red 
and green channels were originally paired.

I gave similar advice to another group some years back, see

Missing channels in two-colour microarray experiments: combining 
single-channel and two-channel data.
Lynch AG, Neal DE, Kelly JD, Burtt GJ, Thorne NP.
BMC Bioinformatics. 2007 Jan 25;8:26.

The above approach using duplicateCorrelation() requires no special 
programming, but has the same effect.

Best wishes
Gordon

> Date: Thu, 11 Nov 2010 17:05:55 +0100
> From: "Edwin Groot" <edwin.groot at biologie.uni-freiburg.de>
> To: bioconductor at stat.math.ethz.ch
> Subject: [BioC] Limma: How to analyze 1- and 2- colour Agilent
>
> Hello all,
> I have a special Agilent 4x44 K microarray analysis problem.
> There are 4 replicates of 2-colour arrays comparing two cell types.
> I want to add to this 4 replicates each of 3 other cell types, but they
> are 1-colour hybridizations (array design is the same).
> How should I do the preprocessing?
>
> My instinct is to normalizeWithinArrays() the 2-colour data, then
> coerce these to single-channel (an EList?). Next background-subtract
> the 1-colour data into an EList. Next, combine these two objects and
> normalizeBetweenArrays(), if necessary. Finally, model and extract
> contrasts.
>
> The alternative was to fool Limma into reading red and green channels
> of the 2-colour data into an EListRaw, combining with the 1-colour
> data, and normalizeBetweenArrays(). However, that omits the
> normalizeWithinArrays().
>
> TIA,
> Edwin
>
> FWIW:
> sessionInfo()
> R version 2.11.1 (2010-05-31)
> i486-pc-linux-gnu
>
> locale:
> [1] LC_CTYPE=en_US.UTF-8       LC_NUMERIC=C
> [3] LC_TIME=en_US.UTF-8        LC_COLLATE=en_US.UTF-8
> [5] LC_MONETARY=C              LC_MESSAGES=en_US.UTF-8
> [7] LC_PAPER=en_US.UTF-8       LC_NAME=C
> [9] LC_ADDRESS=C               LC_TELEPHONE=C
> [11] LC_MEASUREMENT=en_US.UTF-8 LC_IDENTIFICATION=C
>
> attached base packages:
> [1] stats     graphics  grDevices utils     datasets  methods   base
>
>
> other attached packages:
> [1] limma_3.4.2
>
> loaded via a namespace (and not attached):
> [1] tools_2.11.1
>
> Dr. Edwin Groot, postdoctoral associate
> AG Laux
> Institut fuer Biologie III
> Schaenzlestr. 1
> 79104 Freiburg, Deutschland
> +49 761-2032945

______________________________________________________________________
The information in this email is confidential and intend...{{dropped:4}}