[BioC] Intensity distribution of "significant genes"

[Laurent Gautier]

On Wed, Jan 21, 2004 at 01:53:43PM -0500, Phguardiol at aol.com wrote:
> Hi all,
> this a pretty general question ...
> let say I have identified a set of genes differentially expressed between 
> group A and group B and I m using Affy chips (but this could be also interesting 
> for 2 color arrays). 
> Is there a way - something already available) to see the distribution of 
> these down and up-regulated genes (with 2 different colors) on the chip slides ? 
> In addition, if we have "the impression" that the 2D distribution of these 
> spots is not random with, for instance, a high density of these in the "left up 
> corner" or somewhere else, is there a statistical test - so that it is no more 
> an impression - to say that in some parts of the image, the distribution of 
> these differentially expressed genes does not seem to be random suggesting a 
> problem with the procedure or the slides ?

Spatial statistics have certainly nifty things to use (point process analysis
for example ?)

> I think that looking at the slides is 
> necessary but not enought... ?

Well, I would be believe that something obvious on a figure is meaningful...

> Hoping that all this is understandable...
>From your mail, I understand that you have the plots already (and do know
about 'plotLocation', 'indices2xy' or friends).

Explanations for the phenomenon can be multitple, and you are probably 
the best person to tell since you ran the experiments (or are in contact
with the people who did). To give you ideas, here are examples of explanations: 
- chip not properly set during scanning (causing the confocal microscopes
to be slightly out-of-focus a systematic way)
- storage of the chips (say they are all piled with the same orientation,
with one corner is close to heat/light/mobile phone... it has been seen
with cDNA arrays) 
- the way the chips are handled during the process (finger pressing for example)


Note that Affymetrix randomized the probe pairs across the chips because
of such things (the Hu6800 for example was designed before that, and
all the probes in a probe set are lined up)



Hoping it helps,




L. 


> Thanks
> Philippe
> 
> 	[[alternative HTML version deleted]]
> 
> _______________________________________________
> Bioconductor mailing list
> Bioconductor at stat.math.ethz.ch
> https://www.stat.math.ethz.ch/mailman/listinfo/bioconductor

-- 
--------------------------------------------------------------
Laurent Gautier			CBS, Building 208, DTU
PhD. Student			DK-2800 Lyngby,Denmark	
tel: +45 45 25 24 89		http://www.cbs.dtu.dk/laurent