Entering edit mode
Hi Alexandra,
This is a common question, so I'm copying the Bioconductor mailing
list.
The background log-2 expression for SCAN is centered at 0 with a
standard deviation around 1. This means that that ~95% of the probes
measuring background (~80% on the ST array) should have values between
-2 and 2. Then the probes measuring signal should be in the 1+ range.
So what you should see is a right skewed distribution with values
between -2 and say 4 or 5. This is totally expected and normal
behavior for SCAN.
Furthermore, negative values mean that the probe's expression was
below the expected background estimate. Likely this means that the
probe is only measuring background noise and not relevant gene
expression.
Regards,
-Steve
From: Alexandra Muñoz <abm362@nyu.edu<mailto:abm362@nyu.edu>>
Date: Thursday, October 11, 2012 Thu, Oct 11, 2011 1:28 PM
To: Stephen Piccolo
<stephen.piccolo@hsc.utah.edu<mailto:stephen.piccolo@hsc.utah.edu>>
Subject: Expression Values for SCAN.UPC?
Dear Stephen,
I am working with the SCAN function in your package and have a few
questions that I would appreciate if you could answer. I am concerned
about the format of the expression values I have generated and am
unclear as to whether I am using the function correctly.
All of my expression values for samples normalized with this function
are highly compressed between the range of -1-3 approximately. Is
this normal for your process? What type of expression values does your
function produce(log2,log10,..)? Why are there negative values?
I have included the script I used as well as expression values
generated using RMA and the SCAN function. These were performed on
an Affymetrix 1.0 ST array.
Thank you for your time.
Best,
Alexandra
[[alternative HTML version deleted]]