[BioC] Agilent one colour

[Sean Davis]

On Tue, Mar 23, 2010 at 12:47 AM, Yogi Sundaravadanam
<yogi.sundaravadanam at agrf.org.au> wrote:
> Hi Sean,
>
> Thanks for this.
>
> A bit off topic, but with two colour agilent array, I have been using just the LogRatio for downstream analysis. It makes me wonder if I should be normalising this as well.
>

Hi, Yogi.  You might need to look into some background on microarray
data analysis.  Generally, if you want to compare between microarray
experiments, you'll want to do a significant amount of quality control
and assessment to be sure and I think most people would agree that
normalization usually improves downstream inference.  In other words,
QA/QC and normalization are probably necessary steps in microarray
data analysis.

> Sorry, but I have never worked with Agilent, and FE doesn't say much.

The Agilent FE manual does not lay out an analysis recipe, I agree.
However, it is useful for determining what data are represented in the
FE .txt file.  I would recommend having it handy when analyzing
Agilent data, in any case.

Sean


> -----Original Message-----
> From: Sean Davis [mailto:seandavi at gmail.com]
> Sent: Tuesday, 23 March 2010 10:38 AM
> To: Yogi Sundaravadanam
> Cc: bioconductor at stat.math.ethz.ch
> Subject: Re: [BioC] Agilent one colour
>
> On Mon, Mar 22, 2010 at 6:23 PM, Yogi Sundaravadanam
> <yogi.sundaravadanam at agrf.org.au> wrote:
>> Hi all,
>>
>> I am working with Agilent one-colour array, and I find working with gProcessedSignal to be just horrible. I was just wondering if people have had experiences normalising one-colour array, and what works best.
>>
>>
>> I was wondering if Background subtraction, followed by quantile normalisation is preferable?
>
> Hi, Yogi.
>
> gProcessedSignal is background-subtracted already, I believe, but it
> is not normalized.  Very rarely does the data from a one-color array
> platform come normalized between arrays, a necessary step, I think.
> Quantile normalization between arrays is a reasonable possibility,
> yes.  As with any array study, you'll want to do some quality
> assessment both before and after normalization.
>
> Sean
>