[BioC] CEL files with different size

[Joern Toedling]

Hello,

once the format issue is solved, you should thoroughly assess whether it
is reasonable to normalize the samples from the two facilities together,
i.e. how strong and of what nature the batch (facility) effect on the
scanned intensities is. For example, the package arrayQualityMetrics,
which is part of the current development version and of the upcoming
Bioconductor release 2.1, is a straightforward way to produce a number
or plots that can help you to assess these differences and the quality
of your arrays in general. In general, boxplots of the intensities
before and after normalization are a good starting point and even
better, in case you have biological replicates across both batches,
would be scatter plots between those two or more replicate samples. Some
batch effects are nicely resolved using normalization, more persistent
ones have to be taken into account during later data analysis and
explicitly modeled, in those cases it could actually be more reasonable
to normalize each batch of samples separately and then combine their
ExpressionSet objects for the actual analysis. Have a look at limma's
user guide PDF and the sections about factorial designs therein to get
an idea how such models could look like. How to accommodate such a batch
effect really depends very much on its nature and its degree, though. I
would really suggest that you investigate how strong the batch effect is
and if it might pose an issue for normalization and maybe get back to
the list with a description of it.

Best regards,
Joern

Henrik Bengtsson wrote:
> Hi <no name>,
>
> if they are from the same chip type, the smaller ones are probably in
> a binary format and the larger ones in an ASCII format.  Since it is
> much faster to work with binary files, I recommend you to convert the
> ASCII files into a binary files.  You can use convertCel() of the
> affxparser package for this.
>
> The question of normalizing for lab effects is a different issue, and
> I'll leave that for someone else to comment on...
>
> /Henrik
>
> On 10/1/07, apolyzos at bioacademy.gr <apolyzos at bioacademy.gr> wrote:
>   
>> Hello,
>> having conducted 2 microarray experiments of the same cell line, and on
>> the same affy chip in two seperate facilities we were given CEL files of
>> different size (13 & 30 MB). Can I compare them? Is there a way to
>> normalize them together?
>>