Question

Affymetrix Microarray Between-array normalization

1

Entering edit mode

4.9 years ago

va90 ▴ 50

Hi all,
I would be grateful if someone could help me with my question.

I have downloaded GEO Series Matrix data using GEOquery. When I look at box plots and density plots it looks my data needs between-array normalization which I want to do by Limma using "normalizeBetweenArrays".
I wanted to know when should we do that?
Is it better to do between-array normalization before the assignment of probes to genes (when we still have control and many other probes which do not assign to any genes) or is it better to do that after assignment of probs to their corresponding genes and probe-level summarization?

Bests,
Vahid.

R • 1.4k views

ADD COMMENT • link 4.9 years ago by va90 ▴ 50

0

Entering edit mode

Hi Vahid, the data is likely already normalised by RMA, which adjusts for between-array effects via quantile normalisation. normalizeBetweenArrays() is usually used for Agilent arrays.

Which GEO accession are you using? There is usually a blue button toward the bottom of the main accession page, like this:

Click on that, proceed to the R Script tab, and then use the code that is displayed for the purposes of directly reading the data into R. It should already be normalised.

ADD REPLY • link 4.9 years ago by Kevin Blighe 87k

0

Entering edit mode

Hi Kevin,
Thank you so much for your response,
I am using this data "GSE61723". I have already downloaded the data using the method you mentioned. But if you see these plots
click here, click here, click here, click here the distribution does not look similar. In this case (The data already processed through RMA), can we do another between array normalization?

BTW, if we need to do the between array normalization, Is it better to do between-array normalization before the assignment of probes to genes (when we still have control and many other probes which do not assign to any genes) or is it better to do that after assignment of probs to their corresponding genes and probe-level summarization?

Thanks

ADD REPLY • link 4.9 years ago by va90 ▴ 50

0

Entering edit mode

Hey, look at the plots, it looks like the normalisation was successful. In the box-and-whisker plots, those outliers are expected. I am not sure of the validity of performing a second normalisation - that may eliminate the information that you are aiming to detect via your statistical tests.

I am not sure of the answer to your other question. The control probes are used for background correction. After this, the between-array normalisation can occur.