Caimiao Wei | 1 Oct 01:06 2004

Software and references for SNP chip data analysis

 Does anyone know what software or packages are available for SNP chip data 
pre-processing and analysis in addition to dChip by Li and Wong?

Thanks,

Caimiao
Jeremy Gollub | 1 Oct 07:47 2004
Picon

marrayLayout difficulties

Hi, all -

I'm experiencing very poor performance using the marray package (20
minutes to normalize a single <32,000 spot microarray).  Can someone
tell me whether this is normal, or what I'm doing wrong?

In the process of hunting down some errors, I also noticed some odd (to
me) behavior in the marrayLayout maSub slot assignment method, described
below.  An attempt to "correct" this results in a much faster
normalization (~1 minute), which looks good according to the MA plot
but produces different numbers in maM than the slower calculation.

It seems unlikely that either result is correct (I can choose between
suspiciously bad performance, or messing with the marrayLayout object's
internals).

Thanks for any suggestions - details follow.

I'm using R version 1.9.0 on a sparc system running Solaris 2.9.  My
marray version is 1.5.14.

I have a text file, "dat.txt," containing the data I want to normalize.
10 columns, all numeric: in order,
	FEATURE		spot number 1 - 31736
	SECTOR		unnecessary and unused
	ROW		"
	COL		"
	PLATE		ID of printing plate
	Gf		green channel foreground
	Rf		red channel foreground
(Continue reading)

Jean Yee Hwa Yang | 1 Oct 09:52 2004
Picon

Re: marrayLayout difficulties

Hi Jeremy,

That sounds very slow from my experience.  Which image analysis software
did you get your data from?  If you send me an example file off-line, I
will take a look at it for you, I need to take a look to see if maSub was
set properly, as this does make a big different in print-tip
normalization.

Alternatively, try the latest verion 1.5.17 that is temporary place at
http://arrays.ucsf.edu/software/

maNorm was previously very slow for global lowess normalization for larget
number of spots but in the new version, we have speed up the code with
sampling.  However, I don't think this was your problem.

I will also suggest trying the swirl data within the marray package and
see how long that take on yoru computer

data(swirl)
norm <- maNorm(swirl)

If that takes a min or so that there is something wrong with your data
setup.

Cheers

Jean

On Thu, 30 Sep 2004, Jeremy Gollub wrote:

(Continue reading)

michael watson (IAH-C | 1 Oct 13:35 2004
Picon
Picon

Extracting Single Channel Intensities in limma

Hi

I have performed print-tip loess normalisation followed by quantile
normalisation in limma.  I have then converted the resulting MAList
object back to an RGList object using RG.MA().  I now want to extract
the single channel intensities from this normalised RGList object to
create *new* RGList objects which represent comparisons between samples
that were not made on the arrays directly.

However, as RG.MA gives me log2(intensity) back rather than raw
intensity, I presume I need to un-log this data before sending it down
my normal analysis pipeline of RGList -> MAList -> lmFit(MAList) ->
eBayes() -> topTable() ??

Thanks
Mick
Sean Davis | 1 Oct 13:52 2004
Picon

Re: Extracting Single Channel Intensities in limma

Mick,

Is there no way to do this with an appropriate contrast matrix?  That 
seems a safer/more appropriate way of using your two-channel data.

Sean

On Oct 1, 2004, at 7:35 AM, michael watson (IAH-C) wrote:

> Hi
>
> I have performed print-tip loess normalisation followed by quantile
> normalisation in limma.  I have then converted the resulting MAList
> object back to an RGList object using RG.MA().  I now want to extract
> the single channel intensities from this normalised RGList object to
> create *new* RGList objects which represent comparisons between samples
> that were not made on the arrays directly.
>
> However, as RG.MA gives me log2(intensity) back rather than raw
> intensity, I presume I need to un-log this data before sending it down
> my normal analysis pipeline of RGList -> MAList -> lmFit(MAList) ->
> eBayes() -> topTable() ??
>
> Thanks
> Mick
>
> _______________________________________________
> Bioconductor mailing list
> Bioconductor@...
> https://stat.ethz.ch/mailman/listinfo/bioconductor
(Continue reading)

Andrew Beckerman | 1 Oct 14:06 2004
Picon

Re: Rgraphviz install issues

Jeff - thanks for the advice, but unfortunately, no dice.... I download  
the development Rgraphviz pacakge, and using the same routine with  
exporting my library location for graphviz, get the same error.... any  
more ideas?  the offending line seems to be this: ld: can't locate file  
for: -ldotneato

cheers
andrew

dyn092203:~ apb$ export LD_LIBRARY_PATH=/opt/local/lib/graphviz
dyn092203:~ apb$ sudo R CMD INSTALL /Users/apb/Rgraphviz_1-1.4.23.tar.gz
Password:
* Installing *source* package 'Rgraphviz' ...
checking for graphviz... checking for dotneato-config...  
/opt/local/bin//dotneato-config
/opt/local/bin//dotneato-config
configure: creating ./config.status
config.status: creating src/Makevars
** libs
gcc -no-cpp-precomp -I/Library/Frameworks/R.framework/Resources/include  
-I/opt/local/include/graphviz  -I/usr/local/include  -Wall -fno-common   
-g -O2 -c Rgraphviz.c -o Rgraphviz.o
In file included from /opt/local/include/graphviz/render.h:45,
                  from common.h:22,
                  from Rgraphviz.c:1:
/opt/local/include/graphviz/macros.h:34:1: warning: "NEW" redefined
In file included from common.h:13,
                  from Rgraphviz.c:1:
/Library/Frameworks/R.framework/Resources/include/Rdefines.h:129:1:  
warning: this is the location of the previous definition
(Continue reading)

Sean Davis | 1 Oct 15:07 2004
Picon

Re: Extracting Single Channel Intensities in limma


On Oct 1, 2004, at 8:50 AM, michael watson (IAH-C) wrote:

> You tell me, contrast matrices are a black art when it comes to
> complicated experiments.
>
> What I have are two conditions, +ve and -ve, and a control.  The design
> of the experiment is rather odd.  I have arrays that are:
>
> -ve / +ve
> -ve / +ve
> -ve / +ve
> Etc
> Etc
> -ve / Control
>
> What I want are:
>
> -ve / Control
> -ve / Control
> Etc
> +ve / Control
> +ve / Control
> Etc
>
> I figured that by using the methods decsribe in limma, I could subtract
> the single channel intensity data, and completely re-arrange the
> experiment such that all of my -ve and +ve values have the control as
> the denominator.
>
(Continue reading)

michael watson (IAH-C | 1 Oct 15:22 2004
Picon
Picon

RE: Extracting Single Channel Intensities in limma

I'm not sure I told you enough about the experimental design to get this
right, so if we want to explore this using contrast matrices, I better
add the little detail that might make things more complicated.  

Although I can assume all my negative samples are replicates of the same
thing, I can't assume that all my positive samples are.  It's a *very*
strange experiment, but I have 8 +ve phenotype samples and they are NOT
replicates of one another.  So I actually have a factor with nine levels
(the 8 +ve phenotypes and the control) all against a common reference,
which is the -ve phenotype.  To complicate things further, sometimes the
-ve is labelled Cy5, sometimes Cy3, and only two of the +ve phenotypes
have replication, in the form of a single dye-flip.

This is why I really wanted to extract the single channels and then
construct "dummy" microarray experiments ;-)

-----Original Message-----
From: Sean Davis [mailto:sdavis2@...] 
Sent: 01 October 2004 14:08
To: michael watson (IAH-C)
Cc: Bioconductor
Subject: Re: [BioC] Extracting Single Channel Intensities in limma

On Oct 1, 2004, at 8:50 AM, michael watson (IAH-C) wrote:

> You tell me, contrast matrices are a black art when it comes to 
> complicated experiments.
>
> What I have are two conditions, +ve and -ve, and a control.  The 
> design of the experiment is rather odd.  I have arrays that are:
(Continue reading)

Matthew Hannah | 1 Oct 15:38 2004
Picon

Limma p-value distributions, false +ve/-ve etc...

Hi,

I guess the simple question is would you expect or have you seen a
'standard' distribution of p-values for a treated-untreated comparison
(3 reps) after the eBayes procedure in Limma?

Expressed in my usual 'comprehensive' style ;-) 
Following on from a previous thread I've started to look more into the
p.value distributions to get an idea of false +ve and -ve rates. I
understand that p-values should be approx. uniformly distributed as they
approach 1. The following paper uses a mixture of beta and uniform
distributions to model false/true +ve and -ve rates.
http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?cmd=Retrieve&db=PubMed&lis
t_uids=12835267&dopt=Abstract
Code that can be pasted into R is found here -
http://www.stjuderesearch.org/depts/biostats/BUM/index.html

Has anyone done similar? And is this approach valid using the eBayes
moderated stats of limma?

This approach 'appears' ok if you have alot of replicates (my 9
genotypes x 2 treatments x 3 reps example again) ie: the p-values show
the expected distribution. However, if you drop down to a single
genotype and therefore a 3 x 3 comparison the p-values aren't well
distributed (slightly more 0.55-0.8, slightly less 0.85-1). I'm worried
this means that maybe the tests assumptions aren't met, but is there a
way of formally testing this? At the same time I suppose it's not too
suprising that with low replications the distribution is not ideal -
hence my question of other peoples experience with p-value distributions
from limma). Incidently limma p-values have a better distribution than a
(Continue reading)

Jeff Gentry | 1 Oct 16:29 2004
Picon

Re: Rgraphviz install issues

> Jeff - thanks for the advice, but unfortunately, no dice.... I download  
> the development Rgraphviz pacakge, and using the same routine with  
> exporting my library location for graphviz, get the same error.... any  
> more ideas?  the offending line seems to be this: ld: can't locate file  
> for: -ldotneato

That doesn't appear to be the same error at all, unless I'm missing
something.  That error means that you haven't pointed your LD_LIBRARY_PATH
to where the graphviz libraries are sitting.

Gmane