[BioC] Re: how to evaluate the arrays quality?

[David Kipling]

Hi

Also have a look at the simpleaffy package, which has some familiar but
useful QC criteria presented in a coherent fashion.  My own generic QC
pipeline goes something like:

require(affy)
require(affyPLM)
require(simpleaffy)

#    Input the cel files to an AffyBatch object
data1 <- ReadAffy()

#    simpleaffy QC (scaling factors, 3¹/5¹ ratios etc)
qcdat <- qc(data1)
plot(qcdat)

#    Look at the distribution of the PM values
boxplot(data1)
 
#    Look at some RNA degradation plots
deg <- AffyRNAdeg(data1)
plotAffyRNAdeg(deg)
plot(summaryAffyRNAdeg(deg)[1,])

#    Do some probe level modelling
Pset <- fitPLM(data1)
RLE(Pset)
NUSE(Pset)
image(Pset, type=²resids²)


All this is covered in the affy, affyPLM and simpleaffy vignettes.   That
being said, RMA and MAS5 are pretty robust against slightly dodgy chips (too
bright, too dim, fluff, scratches, etc) but what they can't cope with is
duff RNA.  RNA degradation will affect different probesets differently
because of their relative location to the polyA tail, and no amount of
flashy post-image processing or normalisation will rescue that.

David

R 2.1, BioC 1.6

> Dear Han,
> 
>> before starting any analysis on Affymetrix arrays is important to evaluate
>> the arrays quality.
>> How to evaluate the Affymetrix Genechips  image artifacts is ok, ugly or
>> bad but still reasnable >using  using the fitPLM of the affyPLM package?
> 
>> How to write the scripts?
> 
> About the package affyPLM, I would recomend:
> 1.- Read the vignette:
> affyPLM: Model Based QC Assessment of Affymetrix GeneChips
> 2.- Go to the Ben Bolstad web page, where you can see useful images of
> artifacts and "bad" chips:
> http://www.stat.berkeley.edu/~bolstad/PLMImageGallery/
> 
> About the arrays quality, I would recommend to do some quality controls of
> the Affy package (boxplots, histograms, images of the chip, RNA degradation
> plots, etc.)
> 
> HTH
> 
> Jordi Altirriba
> 
> _______________________________________________
> Bioconductor mailing list
> Bioconductor at stat.math.ethz.ch
> https://stat.ethz.ch/mailman/listinfo/bioconductor
> 

Professor David Kipling
Department of Pathology
School of Medicine
Cardiff University
Heath Park
Cardiff CF14 4XN

Tel:    +44 29 2074 4847
Fax:    +44 29 2074 4276
Email:  KiplingD at cardiff.ac.uk