J-Express forum

J-Express => Forum => Topic started by: kathrinr on April 18, 2008, 11:47:44 PM



Title: Normalization algorithms for one-channel Agilent data
Post by: kathrinr on April 18, 2008, 11:47:44 PM
Hello
I'm normalizing one-chanel Agilent data using, Create Sub Data Set and chosing High Level Mean Normalization, High Level Variance Normalization or High Level Mean and Variance Normalization. I am not able to find a presise (read formula) description og these algorithms. Where may I find them ?

Regards,
Kathrine


Title: Re: Normalization algorithms for one-channel Agilent data
Post by: Bjarte Dysvik on April 25, 2008, 11:26:13 AM
Hi Kathrine,

The mean normalization normalize all rows (genes or probes) in your dataset do they have the same mean (0.0). The mean and variance normalization normalize them to have the same mean (0.0) and the same variance (1.0).

This is probably not what you want to do with your Agilent data. Choosing which normalization method to use for your input data is an important, but difficult task. You should see if you can get some recommended procedures from your data provider (NMC?).. A possible lead for one channel data is the quantile normalization option in J-Express.

Bjarte


Title: Re: Normalization algorithms for one-channel Agilent data
Post by: kathrinr on April 30, 2008, 08:55:55 PM
Hello,
thank you for useful information.
You're right that this is not what I want to do with my one-channel agilent data.
But these are the only normalization methods besides Quantile-normalization that I was able
to find in J-Express Pro for one channel data without a base-line micorarray.

The normalizationmethods I've found so far is in SpotPixSuite and the ones I asked you about.
Are there more normalization methods in J-Express Pro supporting one-channel data without a
base-line array? If so,I have not managed to fine them.

Regards,
Kathrine


Title: Re: Normalization algorithms for one-channel Agilent data
Post by: Bjarte Dysvik on May 02, 2008, 10:39:20 AM
Hi again Kathrine,

Why can't you use quantile normalization? It should work well with any one-channel microarrays.


Bjarte