[BioC] nimblegen ChIP to chip array

[Sean Davis]

On 2/21/06 17:25, "ping zheng" <zheng_p17 at hotmail.com> wrote:

> Hello everyone,
> 
> I have never analyzed microarray data, I would like to use
> Biochonductor to analyze data provided by nimblegen. Method used was ChIP to
> chip
> array and data is provided in the following format
> 
> *_ratio.gff, *_pair.txt, *_report.tab
> 
> How can I use normalizeBetweenArrays or Loess function for the data
> provided by nimblegen in the format specified above?
> Can someone tell me what type of data is required for applying these
> functions?
> Does the data format vary from company to company like nimblegen and
> affymetrix?

Formats are in general different from company to company.  The gff file is
in gff format (details are described here:
http://www.sanger.ac.uk/Software/formats/GFF/) and includes the ratio of Cy3
to Cy5 (or vice-versa, depending on the settings for the image analysis
software--clarify with Nimblegen if you have questions on this) in the
"score" column.  GFF can be read as a tab-delimited text file.

As for normalization, I don't think that you want to loess normalize
chIPchip data if you can help it.  Remember that loess is going to take
curvature out of the ratio-vs-intensity plot; if you have a good chIP/chip
experiment, there WILL BE curvature in this plot that represents SIGNAL.
The choice of normalization will affect what tests you can rationally
perform and what tests you want to perform may dictate what makes the most
sense for normalization.

As a final comment, if you haven't done any microarray analysis before, I
would suggest finding a collaborator for helping out with the data analysis
for chIP/chip data on the nimblegen platform.  It can get complicated
quickly if you have replicates or different conditions/antibodies.
Complicating things even more is the relationship between chip signal and
gene expression, if you decide to look at that.

Sean