[BioC] affy quality analysis

[Paul Grosu]

Hi Marinus - 

The procedure we follow when doing QC on Affymetrix arrays is the following
with comments (#):

# Loads the affy library, at the same time we set the working directory

> library(affy)  

# Load the data from the working directory

> AffyData = ReadAffy()

# Load the Probe Level Model library to look for problematic areas on the
chip

> library(affyPLM)

# Run and fit the Probe Level Model on the data

> Pset.AffyData = fitPLM(junhao)

# Set the image export to a PDF file or you can use png() to have individual
images exported

> pdf()

# Export the images of each Probe Level Model chip to see where the problem
areas (patches, scratches, dust spots, etc.) are so that they can be masked
out with GCOS

> image(Pset.AffyData)

# Make sure all the image files have been written out

> dev.off()

# Run the RNA degradation function on each chip to see if there is
consistent degradation of the RNA from 5' -> 3' or if there is aberrant
degradation to know which chips to ignore from the analysis.

> AffyData.deg = AffyRNAdeg(AffyData)

# Run the plot function for the RNA degradation to see visually how the
degradation is from 5' -> 3' across the chips

> plotAffyRNAdeg(AffyData.deg)

# Set the image to exported out as a png file

> png()

# Export out the image to a png format

> plotAffyRNAdeg(AffyData.deg)

# Run the histogram function on the original loaded data to see if there are
any problem of saturation on the high end and if there the distribution of
intensities is not too tight or too much shifted to the low end.

> hist(AffyData)

# Run the boxplot to see if the distribution of the chip as a comparison
between the chips in order to identify chip which might skew the
normalization.

> boxplot(AffyData)

# Finishes the saving of images to chips

> dev.off()

# Prints out the chip/sample name with its corresponding number/order that
is was loaded.  This is useful for identifying the slope and p-value in the
details of the RNA degradation results which will be printed afterwards.

> pData(AffyData)

# Prints out the slope and p-value (and a few other statistics) regarding
the RNA degradation analysis for each chip/sample.  These come in the order
they were loaded which is listed by the above command.

> AffyData.deg

# This sets up the output to be printed to a file

> sink(file="pdata_deg.txt")

# This can also be implemented by print(pData(AffyData))

> pData(AffyData)

# This can also be implemented by print(AffyData.deg)

> AffyData.deg

# Finishes writing the file out and will continue printing outputs to the
console.

> sink()


Just as a mention we've implemented the above procedure to be automatically
processed via an easy drag-and-drop visual analysis pipeline for any Harvard
researchers and users.  Below is a link of a screenshot:

http://www.people.fas.harvard.edu/~grosu/images/r_qc.jpg


Hope it helps,
Paul


-----Original Message-----
From: Dansen, Ing. M.C. [mailto:DANSEN at voeding.tno.nl] 
Sent: Friday, April 15, 2005 8:06 AM
To: bioconductor at stat.math.ethz.ch
Subject: [BioC] affy quality analysis

when using affy arrays several normalization tools are available, 
but quality analysis tools are limited,.....especially compared to cdna
arrays
does anyone have some code to do some quality analysis
next to the packages affy and simpleaffy?
 

best regards,

marinus

 

TNO-Quality of Life

Dept. Physiological Sciences

Marinus C. Dansen

Bioinformatician

dansen at voeding.tno.nl

+31 (0)306944610

 

 

 

This e-mail and its contents are subject to the DISCLAIMER at
http://www.tno.nl/disclaimer/email.html
	[[alternative HTML version deleted]]

_______________________________________________
Bioconductor mailing list
Bioconductor at stat.math.ethz.ch
https://stat.ethz.ch/mailman/listinfo/bioconductor