Gordon Smyth | 1 Sep 2004 02:33
Picon
Picon
Favicon

Re:Flagging spots (was: Bioconductor documentation

At 05:28 AM 1/09/2004, you wrote:
>The reason that we want to read in more columns is to create the 
>flags.  Some people think that spots should be flagged if (e.g.) mean(Rf) 
>differs considerably from median(Rf), or the s.d. of one of these measures 
>is large.  Right now, they need to create the flags outside of Bioconductor.

I think you might want something like

myfun <- function(x) {
okred <- abs(x[,"F635 Median"]-x[,"F635 Mean"]) < 50
okgreen <- abs(x[,"F532 Median"]-x[,"F532 Mean"]) < 50
as.numeric(okgreen & okred)
}
RG <- read.maimages(files, source="genepix", wt.fun=myfun)

Then all the "bad" spots will get weight zero which, in limma, is 
equivalent to flagging them out. You can proceed with

RG$printer <- getLayout(RG$genes)
RG <- backgroundCorrect(RG) # gives more correction options
MA <- normalizeWithinArrays(RG)

to do print-tip loess normalization in which the flagged spots have no 
influence on the normalization.

Gordon

>--Naomi
>
>At 09:28 AM 8/31/2004 +1000, you wrote:
(Continue reading)

michael watson (IAH-C | 1 Sep 2004 12:55
Picon
Picon
Favicon

Read.maimages question

Hi

If I read in multiple arrays with the read.maimages() function, and set
up "Flags" as one of my annotation columns, does that mean that each of
my arrays will be associated with it's individual genepix flags?  Or are
the annotations just read in from one of the arrays, and those
annotations then used for all of the arrays in the RGList object?

Thanks

Mick
Sean Davis | 1 Sep 2004 13:23
Picon

Common reference design and intercept in limma

I have a common reference design with a couple of factors.  I was 
planning on using model.matrix to set up the design matrix.  Do I need 
to use an intercept term when using model.matrix?

Thanks,
Sean
Gordon K Smyth | 1 Sep 2004 14:55
Picon
Picon
Favicon

Re: Read.maimages question

> Hi
>
> If I read in multiple arrays with the read.maimages() function, and set
> up "Flags" as one of my annotation columns,

Flags are converted into numeric weights. They would not become an "annotation column".

> does that mean that each of
> my arrays will be associated with it's individual genepix flags?

Yes, of course, the flags are associated with individual spots.

> Or are
> the annotations just read in from one of the arrays, and those
> annotations then used for all of the arrays in the RGList object?

No, why would you thihk that?  That wouldn't make a lot of sense.

Gordon

> Thanks
>
> Mick
Gordon K Smyth | 1 Sep 2004 14:58
Picon
Picon
Favicon

Re: Common reference design and intercept in limma

> I have a common reference design with a couple of factors.  I was
> planning on using model.matrix to set up the design matrix.  Do I need
> to use an intercept term when using model.matrix?

Makes no difference. You can do it either way.

Gordon

> Thanks,
> Sean
Alexandra-mariela Dumitru | 1 Sep 2004 11:08
Picon
Picon

Problems with BioConductor

Dear Madam/Sir,
I am a user of the R programm at the BF2i laboratory in Lyon. We are doing
research on the microarrays, and use BioConductor for analyzing our results. I
have installed the new version of R 1,9,1 and wished to install the new version
of Bioconductor too. The coputer I am working with is a Mac and fulfills all
the demanded caracteristics of the installation. But I have encountered some
problems while downloading the libraries. Libraries like marray and some other
are downloaded but "cannot be installed for they are not in any repository". I
would appreciate if you could help me solve this problem.
Sincerely 
Alexandra DUMITRU
Matthew Hannah | 1 Sep 2004 14:21
Picon
Favicon

RE: GCRMA/RMA bimodal distribution

Thanks for the replies.

I guess the whole gist of this is when you have lots of changes and some
biologically interesting 'bias', how can you be sure it's not due to the
normalisation.

Maybe I'm misunderstanding how these exprs measures work, but my basic
worry was that if you have a bimodal distribution, an intensity in the
middle of the range (that fluctuates a bit between chips) may be more
likely to shifted towards one of the two peaks. In contrast a uni-modal
distribution would tend to stabilise (or even compress?) these
fluctuations? Probably I've got this wrong. Also I've come to realise
this is perhaps more about the 'real' expression levels, than what an
exprs measure does to them (see below).

I initially assumed that other chips should have a similar distribution
after gcrma to the U95A chip with which it was optimised. However, after
reading about its design (all known, characterised genes) it became
obvious that more (the majority?) of the genes are perhaps likely to be
present in a given sample. The more common design (inc. ATH1) is based
on predicted genes + known so you will have a greater chance for 'not
present' - perhaps even enough to give the low intensity peak. Indeed
this is probably supported by the 'wider' distribution at low raw
intensities on these arrays.

So happy(ish) with this point I'm left with the other concern... As the
raw intensity distributions on these arrays look quite different to the
U95A, are the gcrma model assumptions still true? Looking at log2 MM vs.
PM for U95A (cloud with slope near to 1) compared to ATH1 or U133A (Nike
"swoosh" overlaid?) just fuelled these concerns. Any comments? Basically
(Continue reading)

michael watson (IAH-C | 1 Sep 2004 16:07
Picon
Picon
Favicon

Unequally spaced replicates in limma

Hi

As I have varying numbers of replicates, and they are not regularly
spaced on the array, and given that I would like a list of
differentially expressed genes which is averaged over replicates, I
assume the best thing to do is normalise my data, and then average over
replicates in the MAList object, and then pass the averaged data to
lmFit() etc?

Is that right?

Cheers
Mick
Jeff Gentry | 1 Sep 2004 16:15
Picon

Re: Problems with BioConductor

> I am a user of the R programm at the BF2i laboratory in Lyon. We are
> doing research on the microarrays, and use BioConductor for analyzing
> our results. I have installed the new version of R 1,9,1 and wished to
> install the new version of Bioconductor too. The coputer I am working
> with is a Mac and fulfills all the demanded caracteristics of the

A couple of questions ...
- By "installing Bioconductor", I am assuming that you are using
getBioC()?  Can you cut & paste some of the relevant output from your
session?

- I'm not super familiar w/ the Mac style R, so my terminology is probably
incorrect here, but are you using the R-aqua or the unix-style R?  That
could make a big difference.

> the libraries. Libraries like marray and some other are downloaded but
> "cannot be installed for they are not in any repository". I would

I don't currently see a problem w/ the repository, using R-1.9.1, I just
did a BioC install and all went well, including marray.  Could you give
more details?

Thanks
-J
Arne.Muller | 1 Sep 2004 16:29

Friedman test

Hello,

I've a rather statistical question ...

I've an experimental design with 2 factors, batch (3 levels) and dose (5 levels) with 0 to 4 technical
replicates (a dose is missing for a certain batch). I've seen that one batch has consistently lower
variance within doses for most of the genes I've looked at.

I'm performing a 2way anova, and was wondering if this screws up the results, since the assumption of equal
variance is violated (quite a lot, I think!).

Does it make sense to "fake" a random block design by taking the mean of the technical replicates and then run
a friedman test? I've never used this test before, and I don't realy feel happy taking the means without
considering the variance of replicates ...

Any hints and thoughts are appreciated,

	+kind regards,

	Arne

Gmane