| Getting Started with MatArray | Search Reference |
Normalization
The data from microarrays typically present many biases and systematic errors that can be removed by a sensible normalization. The toolbox offers different scripts geared at different type of errors.Geometrical normalization
The red / green ratio can be function of the position of the spot on the slide. For instance, the spots on the right side of the slide can be mostly red, while the spots on the left side of the slide are green. It is very likely that such effects are an error and not a genuine signal. The function
normgen can be used to
remove such type of error. Please see the reference for details.
Non-linear normalization
The red / green ratio can be function of the spot intensity. For instance, the brighter spots could be mostly red and the dimmer spots mostly green. It is very likely that such effects are an error and not a genuine signal. The function
normnonlin can
be used to remove such type of error. Please see the reference
for details.
All-in-one
As normalization is a very common process, a do-it-all script,normgen,
has been written. The idea is to hardwire the usual parameters
once and for all.
One of the point of the script is to determine automatically
the geometrical layout of the slide from the number of spots.
For instance, NEN slides can come in 2400 genes, 4800 genes
and 325 genes varieties. The layouts corresponding to those
numbers of genes for NEN slides are hardwired, but if you use
some other slides it is unlikely that the numbers will fit.
Please modify the script to suit your needs (do not be afraid,
it is a simple if...then at the beginning of the script).