[BioC] Identification of unannotated transcripts using Arabidopsis 1.0R Tiling Arrays

[Wolfgang Huber]

Dear Pia,

to supplement Sean's suggestion, you could look at the two papers cited
here: http://www.ebi.ac.uk/huber-srv/queryGene
The analysis for this project was done with the tilingArray package. But
applying this to your data will require a substantial amount of your own
programming, due to the variety of tasks, the tilingArray package does
not provide an easy to use end-to-end it solution; just some buiding
blocks. Still, of course, I hope that it is useful, let me know if you
have questions.

 Best wishes
 Wolfgang


------------------------------------------------------------------
Wolfgang Huber  EBI/EMBL  Cambridge UK  http://www.ebi.ac.uk/huber


Sean Davis ha scritto:
> Pia Sappl wrote:
>> Dear Tiling Array Users
>>
>> We have recently performed an Arabidopsis Tiling Array experiment to 
>> assess differential expression between mock treated and dex treated 
>> samples at 4 and 12 h after dex treatment (2 biological paired 
>> replicates = 8 chips in total).  ie mock 4h versus dex 4 h and mock 12 h 
>> vs dex 12 h.  As part of our analysis we would like to identify regions 
>> of DNA that do not fall within annotated exons (ie unannotated genes).  
>> Is anyone aware of an existing R algorithm we could use (or perhaps with 
>> minimal modifications) that would do this?  So far we have used TAS 
>> (Affymetrix software) to normalise the data and define regions of 
>> positive expression.  We have also used IGB (Affymetrix genome browser) 
>> to manually inspect the results.  However, we would like to take a more 
>> automated and objective approach using R.  Finally, has anyone used the 
>> Tiling Array to identify differential expression? And if so, what 
>> approach was used?
>>   
> I haven't used the package in a long time, but you might look at the 
> tilingArray package.  Also, I would highly recommend reading some of the 
> literature utilizing tiling arrays.  The analysis of such arrays is full 
> of challenges and will probably require some customized analyses to 
> answer specific biologic questions. 
>